ISSN 1055- 1425
November 2005
This work was performed as part of the California PATH Program of the
University of California, in cooperation with the State of California Business,
Transportation, and Housing Agency, Department of Transportation, and the
United States Department of Transportation, Federal Highway Administration.
The contents of this report reflect the views of the authors who are responsible
for the facts and the accuracy of the data presented herein. The contents do not
necessarily reflect the official views or policies of the State of California. This
report does not constitute a standard, specification, or regulation.
Final Report for Toyota GapAdvise
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
Investigation of Elderly Driver Safety and
Comfort: In- Vehicle Intersection “ Gap
Acceptance Advisor” and Identifying
Older Driver Needs
UCB- ITS- PRR- 2005- 36
California PATH Research Report
Benedicte Bougler et al.
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
Toyota GapAdvise:
INVESTIGATION OF ELDERLY DRIVER
SAFETY AND COMFORT:
In- Vehicle Intersection “ Gap Acceptance Advisor”
and Identifying Older Driver Needs
FINAL REPORT
Benedicte Bougler
Delphine Cody
Judy Geyer
Jedidiah H. Horne
James A. Misener
Christopher Nowakowski
Caroline J. Rodier, Ph. D.
David Ragland, Ph. D, MPH
Susan A. Shaheen, Ph. D.
University of California PATH Program
1357 S. 46th Street Bldg 452; Richmond, CA 94804- 4648
Joy Caguimbaga
Bevin Daniels
Kathryn Hamel, PhD
Movement Analysis Laboratory, Department of Physical Therapy and Rehabilitation Science
University of California San Francisco, Box 0625; San Francisco, CA 94143
March 28, 2005
1
ACKNOWLEDGMENTS
The authors would like to thank Toyota Motor Corporation for their generous contributions to
this older driver research.
We would also like to express appreciation to our older driver research partners who supported
our focus group research, particularly: Albert Austria of Toyota Technical Center, USA; Sharon
Smith of the Rossmoor Senior Adult Community in Walnut Creek;
California PATH and Institute of Transportation Studies, Berkeley faculty, staff, and students
also deserve special credit for their assistance with this project, including: Dr. Samer Madanat,
Steven Campbell, Rachel Finson, Linda Novick, Cynthia McCormick, Amanda Eaken, and
Joanna Mui. Additionally, thanks go to members of the PATH “ Intersection Decision Support”
team who prepared experimental software, drove vehicles and served as flaggers. Particular
thanks in this regard go to Susan Dickey, Ashkan Sharafsaleh, Joel VanderWerf and our flagger
Tracy Shaw.
The contents of this report reflect the views of the authors who are responsible for the facts and
the accuracy of the data presented herein. The contents do not necessarily reflect the official
views or policies of Toyota Motor Corporation. This report does not constitute a standard,
specification, or regulation.
2
TABLE OF CONTENTS
EXECUTIVE SUMMARY………..……………………………………………………..... 3
1.0 INTRODUCTION………..………..…………………………………………………. 13
2.0 DETERMINE EXTENT OF PROBLEM………..…...…………………………...... 14
3.0 CONDUCT FOCUS GROUP AND OBSERVATIONAL ANALYSIS
OF ELDERLY DRIVERS…………………..…..………………………………........ 31
3.1 Focus Group Research……………………………………………………….. 31
3.2 Observational Research……………………………………………………… 42
4.0 CONDUCT DRIVING EXPERIMENTS……………................................................ 60
5.0 RECOMMENDED IN- VEHICLE DESIGN…...…………………………................ 82
APPENDIX A: FOCUS GROUP SUMMARIES
APPENDIX B: OBSERVATIONAL GROUP SUMMARIES
APPENDIX C: DRIVING EXPERIMENT SUMMARIES
3
EXECUTIVE SUMMARY
Our work in Toyota GapAdvise is comprised of two interrelated elements: identify driving task
challenges, and a pilot study on one particular class of decision support system, an intersection
gap advisor. From these elements, we have recommended countermeasures and potential design
guidelines for the elderly driving population in the United States.
We performed our work in the following sequence of technical tasks, each corresponding to a
section heading in this final report:
Determine Extent of Problem ( Task 1). From crash databases and demographic data, we
have determined the projected extent of the problem, extending from past work. From
our synthesis and interpretation of data and publications, we have ranked causal factors.
Conduct Focus Group and Observational Analysis of Elderly Drivers ( Task 2). Through
focus groups and observing elderly drivers in their own vehicles, we have developed an
understanding of the problems faced by elderly drivers.
Conduct Driving Experiments ( Task 3). Using PATH instrumented vehicle and test
intersection at the University of California, Berkeley’s Richmond Field Station facility,
we have performed in- vehicle experiments to characterize driver behaviors.
Recommend In- Vehicle Design ( Task 4). From Tasks 1 – 3, we provide integrated
recommendations, to include engineering constraints and design principles, from Tasks 1
– 3.
Determine Extent of Problem ( Task 1)
The growing number of older drivers presents a special challenge and opportunity for health
professionals and the motor vehicle industry. Over the next few decades, the number of persons
over age sixty- five will increase at least 240%, and the number of persons over eighty- five will
increase by at least 466%. In the meantime, the percent of seniors licensed to drive is increasing
4
steadily. Also, today’s older adults entering into retirement are driving more miles per year than
current retirees, as today’s adults are more accustomed to longer commuting, shopping, and
recreation trips than current retirees experienced in their younger adulthood.
The number of licensed drivers and the average annual miles driven are projected to increase for
all age groups. Older adults deserve special attention by health care professionals and the motor
vehicle industry because driving performance tends to decline with age. Adults age sixty- five
and over have higher collision rates both per mile driven and per licensed driver than adults age
twenty- five to sixty- four. Seniors are also overrepresented in certain crash types, such as
crossing path collisions and those involving right- of- way violations.
Our work outlines the projected growth in the number of older adults, older adult drivers, and the
differences in collision outcomes between adults and older adults. Unless otherwise noted, all
data describe the United States population, its drivers, and their collision rates. Data sources and
analysis methods are explained.
Conduct Focus Group and Observational Analysis of Elderly Drivers ( Task 2)
Impediments and potential solutions to safely extend driving for older travelers were explored in
four focus groups conducted in the summer and fall of 2004 at the Rossmoor Senior Adult
Community in Walnut Creek, California. In total, 20 women and 16 men participated in the four
focus groups, and their ages ranged from 70 to 85 years ( mean age of 78). Driving alone was the
most frequently used travel mode among participants, and they owned vehicles that ranged from
small compact cars to luxury sedans.
The focus group research method allows for detailed, in- depth exploration of relatively new
research areas, but its small, non- random sample limits generalizations to the larger population.
As a result, it is important to interpret the results of the focus group findings in the context of the
demographic and attitudinal profiles of the participants. These were assessed using
questionnaires administered before the start of each focus group. The survey results indicate that
participants in this study were most likely to:
5
· Have begun driving at 18.5 year of age;
· Be married;
· Live in a household with 1.5 people, 1.5 drivers, and 1.4 autos;
· Have a Bachelor’s degree; and
· Have a household income ranging from $ 20,000 to $ 49,000.
In addition, the typical focus group participant expressed the following attitudes related to auto
use:
· Enjoyed and was satisfied with his/ her personal vehicle;
· Did not find operation and maintenance of a personal vehicle to be onerous; and
· Neither inclined nor disinclined to experiment with new things.
During the focus group discussion, several key problem areas were identified. In approximate
order of importance, these included:
· Blind spots while merging and changing lanes, often exacerbated by the difficulty drivers
experienced turning their necks;
· Problems reversing and parallel parking, again caused by blind spots and difficulty
looking backwards;
· Items placed in the trunk not staying in place while driving;
· Seats too low for drivers to see above the dashboard and/ or reach the pedals;
· Difficulty with vehicle ingress and egress, particularly for taller drivers and those with
physical disabilities, and often worsened by poor seating design;
· Problems adjusting or reading knobs, dials, and displays, particularly dim displays and
clocks set into the dashboard at a hard- to- see angle;
· Concern about glare and the speed of oncoming drivers at night or in the rain; and
· Travel to unfamiliar or long- distance destinations.
6
Participants also identified potential solutions to their specific difficulties. Numbers one through
three in parentheses indicate increasing levels of solution complexity.
· Blind spots while merging and changing lanes and concern about the speed of other
vehicles: ( 1) “ wink” mirrors, redesigned convex right hand side mirrors; ( 2) redesigned
window pillars; and ( 3) automated blind spot detection.
· Problems gauging when to safely make left- hand turns at unprotected intersections: ( 3)
intelligent intersections.
· Concern for hitting other cars, the curb, or pedestrians when parallel parking and
reversing: ( 1) reverse beepers ( to avoid hitting other cars or pedestrians), “ curb feelers”
( to avoid hitting the curb); and ( 3) automated parking technology.
· Items not staying in place when placed in the trunk and difficulty lifting items from the
trunk when loading or unloading: ( 1) netting, bungee cords, Velcro; and ( 2) “ flat” trunks
without additional lip, compartmentalization.
· Seats too low for drivers to see above the dashboard and/ or reach the pedals: ( 1) manual
up/ down adjustments on vehicles; and ( 2) electric- adjust memory seat settings, adjustable
pedals.
· Difficulty reading displays and using knobs: ( 2) increased brightness, knobs on steering
wheel, remote for radio.
· Physical discomfort or difficulty during access and egress due to limited range of motion
or physical impairment: ( 1) handles above door, running boards, mechanical door check
to avoid slamming; and ( 2) ergonomic design for taller drivers, adjustable steering
wheels, sliding front doors.
· Decreased visual acuity when driving at night or during rain: ( 2) automatic- dimming
headlights for incoming glare, faster automatic lights for night driving.
· Traveling to unfamiliar locations increases anxiety: ( 1) digital compasses; and ( 3) GPS-enabled
in- vehicle navigation systems ( can also mitigate short- term memory loss).
· Sun glare: ( 1) wider or adjustable visors; and ( 2) tinted windshields.
· Problems remembering when to turn off turn signals: ( 1) volume setting, timeout
function.
7
An important element of this was to observe and analyze older adults during “ in- vehicle”
performance on an open road course and also during ingress/ egress tasks, as it was hypothesized
that problems faced by older drivers would be clearly observed through analysis of “ in- vehicle”
performance. It was also hypothesized that the problems detected in this study would direct
future research on specific intervention strategies to address these problems. Future motor
vehicle modifications, along with medical and behavioral intervention strategies should be
targeted at keeping older drivers safe on the road, despite functional declines. Three components
were a Rossmoor driving section, a Walnut Creek driving section and observation of
ingress/ egress.
Key results from the Rossmoor section include:
· 75% of those drivers who reversed out of the starting parking space did not fully look
through rear window before backing out
· 100% of those who pulled forward out of the parking space made no scanning errors
· Many errors were made during turn out of Rossmoor parking lot:
o 90% did not fully stop before turning
o 43% did not scan the surrounding area adequately
o 20% failed to slow
o 23% failed to signal
· 40% of drivers made head turning errors at stop sign controlled intersections
· 67% of drivers made head turning errors during lane changes
· 17% of drivers made head turning errors during yield
· 13% of drivers made signaling errors at intersections
· 23% of drivers made signaling errors during lane changes
· 57% of drivers did not fully stop at stop sign controlled intersections
· 13% of drivers did not follow prescribed route
· 30% of drivers did not adequately scan
· 37% of drivers sped
· 17% of drivers made critical errors
Key points from Walnut Creek section include:
8
· 73% of drivers made head turning errors at intersections
· 77% of drivers made head turning errors during lane changes
· 20% drivers made head turning errors while parking
· 63% of drivers turned too wide
· 17% of drivers failed to signal at intersections
· 17% of drivers failed to signal before changing lanes
· 23% of drivers failed to signal during parking/ pulling out
· 20% of drivers rolled through stop signs
· 43% of drivers inadequately scanned during drive
· 17% of drivers sped during drive
· 17% failed to have two hands on wheel during all of drive
· One driver performed a self- distracting activity while driving ( looking at map, misses
light turning green)
· 10% of drivers committed critical errors
Key results from ingress/ egress observations include:
Suitcase Loading
· 70% placed the suitcase in the trunk
· 21% placed the suitcase on backseat floor
· 9% placed the suitcase on backseat
Grocery Bag Loading
· 64% placed the groceries in the trunk
· 21% placed the groceries on the backseat floor
· 15% placed the groceries on backseat
Ingress
· 28% had difficulties getting into the driver seat
· 67% had difficulties getting out of the driver seat
9
· 65% had difficulties getting into rear passenger seat
· 91% had difficulties getting out of rear passenger seat
· Required the use of one arm/ hand during ingress - driver seat = 12%, backseat = 32%
· Required the use of one arm/ hand during egress - driver seat = 24%, backseat = 23%
· Required the use of two arms/ hands during ingress - driver seat = one person, backseat =
9%
· Required the use of two arms/ hands during egress - driver seat = 9%, back seat = 14%
Conduct Driving Experiments ( Task 3)
We experiment with an in- vehicle message for a left turn across path / opposite direction
( LTAP/ OD) gap advisor, judging its effectiveness with older drivers ( versus younger drivers).
This work leverages research conducted under the Intersection Decision Support ( IDS) project
and upcoming with the Cooperative Intersection Collision Avoidance System ( CICAS). This
gives rise to the LTAP/ OD display used for Toyota GapAdvise.
This experiment is as follows: the subject vehicle ( SV) – or the vehicle equipped with the
Toyota GapAdvise LTAP/ OD warning system – approaches the intersection. It has a
( permissive) green signal, but there is no left turn arrow or protected cycle, so the driver slows
down to a stop to check if it is safe to make a left turn onto at the intersection. The SV driver
may be older or otherwise not able to easily judge the speed or location of this approaching
traffic, making it hard to decide whether or not to turn. While the SV driver is trying to
determine whether the left turn is safe, other vehicles (“ Principal Other Vehicles” – POV) are
approaching the intersection with the intent of proceeding straight. Therefore, intermittent gaps,
some safe and some not save may be present.
In exploring the concept of an in- vehicle gap advice system, this study addressed the following
four research questions on 20 subjects:
1. What is considered an unsafe gap?
10
2. When should you give the warning to be effective in influencing the drivers’ decisions?
3. How should the warning be given?
4. How effective might the system be in reducing the number of unsafe turns?
We are also able to distinguish between the effectiveness of in- vehicle systems versus an
analogous roadside- mounted system, since we are conducting parallel roadside warning
experiments under the IDS project.
Recommend In- Vehicle Design ( Task 4)
We suggest specific solutions that focus on redesign of vehicle components or on changes that
are already available in some models, such as improved mirrors, minor adjustments to displays
or radios, and mechanical seat adjustments and checks on doors. Participant focus group results
also suggest improvements involving more complicated electronics or major structural changes
to vehicle design fall into the second category, and these include redesign for blind spots, flat
trunks, and automated or electronically adjustable features, among other recommendations. We
also provide a set of solutions which integrate enhanced driver information into automatic
vehicle navigation or alert systems.
Although our sample population of older drivers was relatively robust and most likely higher
functioning than the average population of older adults, most drivers in the study made several
driving errors which could affect safety. Our observational analysis of driving performance
confirm the findings from the focus groups which suggest that blind spots, difficulties changing
lanes, and concerns about hitting objects such as a curb or pedestrian were among the most
important problem areas mentioned by our participants. Recommendations for vehicle
modifications include that might address the reduced neck and torso mobility include: mirror
redesign, increased visibility through pillar and window reconfiguration, back- up beepers and
cameras, and potentially a warning system of some sort to remind individuals to scan
appropriately at intersections and during lane changing.
We were surprised to find that 60% of the individuals in our study had deficits in working
memory given that they all easily passed the cognitive screening test. This suggests that
11
navigation could be beneficial in this population; however this idea must be tempered by the fact
that the majority of participants had mild deficits in directed visual search and half had mild
deficits in divided attention.
From a usability standpoint, we observed that those with mobility problems and taller
individuals had the most difficulty getting into and out of the vehicle, particularly for the rear
passenger seat. Additionally, the smallest women in the study tended to be positioned too close
to the steering wheel and sometimes forced into a more flexed, or forward leaning posture.
Greater seat adjustment capability ( particularly for the height of the seat) might address some of
these limitations. Greater space in the back seat, along with some form of adjustment might
improve an older adult’s ability to perform ingress and egress more easily.
The drivers’ comments on the overall concept of a gap advice system were positive. Almost all
of the drivers commented that such a system could be useful and come in handy at times.
However, unsurprisingly, almost all of the drivers also agreed that the interface would need
much more study and work before being accepted as an in- vehicle system.
The head- down display used for the visual component of the warning was reported as being too
low to be seen, even though it was mounted as high as possible for a head- down display. When
asked to comment on the graphical components of the display, such as the looming no- left- turn
sign or the oncoming vehicle distance to intersection countdown bar, all 20 drivers reported that
they did not glance to the display during their turning maneuver, rather they simply listened for
the warning beep. A few of the drivers expounded on this, stating that their eyes and attention
were focused on the oncoming vehicle throughout its approach, and they did not feel
comfortable taking their eyes off the road.
These and other comments spawn potential design considerations:
1. Integrated DVI design, with specific auditory and visual meaning to intersection left turn
conflicts.
2. Recognition that the infrastructure mounted active sign, in the scanning direction of SV
12
drivers, had particular appeal. This may translate into design guidance of head up, not head
down, display location. More specifically, when making left turns drivers tend to scan the
upper left quadrant of the windshield, in the vicinity of the left side A- pillar. 1 This presents a
visual design placement challenge, perhaps resolved by relying on another channel, e. g.,
auditory.
We recommend that future research include the design and possible deployment of prototype
vehicles incorporating different level solutions for field tests with older drivers. Because of the
high cost and uncertain demand for some technologies, it is possible that the marginal benefits of
component level solutions may be the most cost effective for older drivers. Because many
drivers also had difficulty with merging, another area that deserves future study is merging and
turning behavior, perhaps through a merge assist study with technology development and
interface assessment.
We feel that specific GapAdvise driver interfaces be designed for more comprehensive studies in
the future. Some of the studies, both general observational and with intersections, should be
comprehensively designed. For example, the older adult could also be studied driving during
twilight or night hours. Another interesting study would be to evaluate a prototype vehicle using
the same subjects tested in this study to evaluate how their performance changes in a new
vehicle targeted to older adults.
1 Nowakowski, C. ( 2004). Intermediate summary of IDS ( intersection decision support) field test results. Presented
at the IDS Quarterly Meeting 9/ 26- 9/ 29 in Minneapolis, MN. Berkeley, CA: California PATH.
13
1.0 INTRODUCTION
This work was undertaken in recognition that with the growing numbers of elderly drivers,
particularly with the impending retirement of the bow wave of the " baby boom" generation,
most living in relatively low density suburban environments, the mobility challenges will
increase greatly in coming years. The emerging challenge for millions of older adults will be
to maintain driving mobility in the face of functional decline.
This report describes our work, which includes a multi- disciplinary systems- oriented approach
to develop a pilot study on one particular class of decision support system, an intersection gap
advisor, Toyota GapAdvise. Our work also identified driving task challenges, from we which
suggest countermeasures for the elderly driving population by means of interpretation of focus
groups and observations. From these elements, we have recommended countermeasures and
potential design guidelines.
In short, we have performed the following sequence of technical tasks, each corresponding to a
section heading in this final report:
Determine Extent of Problem ( Task 1). From crash databases and demographic data,
we have determined the projected extent of the problem, extending from past work.
From our synthesis and interpretation of data and publications, we have ranked causal
factors.
Conduct Focus Group and Observational Analysis of Elderly Drivers ( Task 2). Through
focus groups and observing elderly drivers in their own vehicles, we have developed an
understanding of the problems faced by elderly drivers. In areas as: ingress/ egress, and
seating/ control adjustments.
Conduct Driving Experiments ( Task 3). Using PATH instrumented vehicle and test
intersection at the University of California, Berkeley’s Richmond Field Station facility,
14
we have performed in- vehicle experiments to characterize driver behaviors. We note
that we have significantly leveraged our Federal- and Caltrans- sponsored Intersection
Decision Support ( IDS) project to focus on gap acceptance ( versus collision warning)
advisor for older drivers2. This has allowed us to add to the Toyota- sponsored segment,
additional observations on driver acceptance of left turn warnings provided from the
infrastructure versus those provided from a driver- vehicle interface ( DVI).
Recommend In- Vehicle Design ( Task 4). From Tasks 1 – 3 , we provided integrated
recommendations, to include engineering constraints and design principles, from Tasks
1 – 3.
2.0 DETERMINE EXTENT OF PROBLEM
2.1 A Growing Senior Population
Traffic safety is an important issue for all segments of the population. Population changes
affect the both the number of motor vehicle passengers and licensed drivers. Also, an increase
in population leads to an increase in motor vehicle injuries and fatalities. The general
population is expected to grow 157% from 1990 to 2040 ( Table 2- 1). These projections,
published by the U. S. Census Bureau, are based on the 2000 Census. 3
Table 2- 1. Projection of U. S. Population
Total Population
1990 249,622,814
2000 282,125,000
2010 308,936,000
2020 335,805,000
2030 363,584,000
2040 391,946,000
The number of older adults in the United States is accelerating not only due to overall
2 The clear distinction is that our approach for Toyota GapAdvise focuses on an in- vehicle gap advisor and elderly
drivers, whereas our Federal IDS project does not address in- vehicle systems, nor does it particularly focus on
elderly drivers.
3 U. S. Interim Population Projections, Based on Census 2000. U. S. Census Bureau, Population Division,
Population Projections Branch. March 18, 2004. http:// www. census. gov/ ipc/ www/ usinterimproj/
15
population growth, but also because of the aging “ Baby Boom” generation and an increasing
life expectancy. In 1990, 12.5% of the population was sixty- five years old and older. This
percentage is expected to increase to 20.4% by the year 2040. Therefore, the senior population
will not only increase, but it will become a more visible demographic group. Seniors eight-five
years and older, a demographic group especially influencing the demands on health and
care facilities, is projected to grow from 1.1% of the population in 1990, to 3.9% of the
population in 2040.
Table 2- 2. Projection of U. S. Senior Population
65+ 85+
1990 31,242,000 2,830,000
2000 35,061,000 4,267,000
2010 40,243,000 6,123,000
2020 54,632,000 7,269,000
2030 71,453,000 9,603,000
2040 80,049,000 15,409,000
The ratio of males to females changes drastically as a function of age, and this change is
important to understanding the needs of the average older driver and passenger. In 1990,
59.8% of the population over the age of sixty- five was female and 72.1% of the population
eighty- five and over was female. Population projections published by the U. S. Census
illustrate an expectation that the average life- span of males and females will increase. Females
may still have longer life- expectancies, but the percent of the 65+ and 85+ populations that are
female will decrease slightly because both men and women are expected to live longer.
Figures 2- 1 and 2- 2 illustrate this expectation:
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
45,000
Population in
1000s
1990 2000 2010 2020 2030 2040
Male
Female
16
Figure 2- 1. U. S. Population Projections, Age 65 and Over
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
Population in
1000s
1990 2000 2010 2020 2030 2040
Male
Female
Figure 2- 2. U. S. Population Projections, Age 85 and Over*
The large increase in the elderly population will bring a substantial increase in demand for safe
mobility for seniors. Currently, middle- aged adults drive more than the current elderly did
when they were younger. Now, adults drive farther distances to work, for errands, and for
recreational purposes than any other generation of adults. The transportation infrastructure
and urban design will not change drastically in the next forty years, so we can expect private
motor vehicle travel to continue to be the most popular form of travel.
17
Implications for GapAdvise
The very substantial increase in older adults ( over 65 and over 85) will mean dramatic
increases in need for mobility. A very substantial proportion of this mobility will be delivered
by the private automobile. Automobile design will need to be modified to meet the demand for
safety and comfort for this elderly population, whether they are drivers or occupants.
2.2 Senior Driver Population
An increase in the senior population will lead to an increase in the number of elderly drivers.
In 1991, 43% of males eight- five and over had a driver’s license. By 2000, 78% in this age
group had a license. Similarly, the percentage of females eighty- five and over who had
licenses increased from 13.5% in 1991 to 36.3% in 2000. Also, the driving patterns of seniors
have changed dramatically in the last 15 years, and will probably continue to change. 4
To our knowledge, there are no published projections of the number of licensed drivers. The
Bureau of Transportation Statistics ( BTS) provides the number of total number of licensed
drivers nationwide by gender and five- year age categories from 1990 to 2001. We can
conclude fairly confidently that the percent of seniors who are licensed drivers will continue to
increase. First, there has been a steady and substantial increase in percent of seniors who are
licensed drivers over at least over the past decade. Second, younger drivers who will be
seniors over the next few decades are more likely to be licensed, and to have driven more,
compared to current seniors when they were younger. However, we do not know the
magnitude and pace of the expected increase. Likewise, we do not know when and if this
increase will level off, aside from the fact that the percent of those licensed in any particular
age group will most likely not exceed the current level of that group.
To produce a projection of the percentage of licensed drivers at each age group in future years,
we have used data on the percentage of licensed seniors from 1990 to 2001 and created
projection models to estimate how this percentage might change between 1990 and 2040. To
4 Rosenbloom, Sandra. The Mobility Needs of Older Americans: Implications for Transportation Reauthorization.
Brookings Institute Series on Transportation Reform. July, 2003.
http:// www. brookings. org/ dybdocroot/ es/ urban/ publications/ 20030807_ Rosenbloom. pdf
Accessed 2/ 20/ 04
18
satisfy the expectation that the percentage will increase and then level off, we have used a
logarithmic regression function to approximate the growth and ultimate leveling- off of the
percentage of licensed drivers. This approach is purely a projection, and the actual percentage
of licensed drivers in each age group will depend on a number of factors, including future
changes in licensing policies, mobility needs based on housing and transportation trends and
policy, and vehicle and highway design. As an example, our projection model for female
licensed drivers age 70 and over is shown in Figure 2- 3.
Female 70+
y = 5.6441Ln( x) + 49
R2 = 0.914
0
10
20
30
40
50
60
70
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Figure 2- 3. Projection Model for Number of Female Licensed Drivers Age 70+*
* Based on the 2000 Census and BTS Licensed Drivers 1990- 2001.
We then multiplied the projected percentage of licensed drivers by the projected population to
obtain the projected number of licensed drivers. Our projections rely on the total population
projections from the 2000 Census. The following two figures ( 2- 4 and 2- 5) show the expected
number of licensed drivers over the age of 70 and 85.
19
0
5,000
10,000
15,000
20,000
25,000
Number of
Drivers in
Thousands
1990 2000 2010 2020 2030 2040
Male
Female
Figure 2- 4. Number of Licensed Drivers Age 70 and Over*
* Based on the 2000 Census and BTS Licensed Drivers 1990- 2001.
0
1,000
2,000
3,000
4,000
5,000
6,000
Number of
Drivers in
Thousands
1990 2000 2010 2020 2030 2040
Male
Female
Figure 2- 5. Number of Licensed Drivers Age 85 and Over*
* Based on the 2000 Census and BTS Licensed Drivers 1990- 2001.
Both figures above illustrate that the number of licensed senior drivers will increase rapidly, at
an even faster rate than the expected increase in the elderly population. Indeed, figures 2- 4
and 2- 5 show a large expected increase from 2000 to 2040 in the number of licensed drivers;
the number of drivers age 70+ and 85+ will increase 252% and 466%, respectfully.
20
Implications for GapAdvise
The increase in the number of older drivers, whether defined 70 and older, or 85 and older, in
conjunction with well established declining function with age, will mean a very substantial
increase in the number of drivers on the nation’s highways with reduced capacity for driving.
There will be a very high, and increasing, demand for altered vehicle design to facilitate safe
and comfortable driving for older drivers.
2.3 Elderly Drivers and Increased Motor Vehicle Injury
Motor vehicle fatality or injury rates are presented in many different ways. Often, a simple
number of injuries are reported. Other times, reports calculate the rate of fatality or injury per
population size, per licensed drivers, or per miles driven. Each method carries different
implications, and they are each discussed here.
The first data analysis method is to study the total number of fatal crashes by age and gender.
These data is often used to provide medical facilities planners and emergency responders
information about the number of crashes, and therefore their agency’s expenditures. The
elderly are involved in far fewer motor vehicle crashes than teenagers and adults, and as a
consequence they suffer fewer injuries and fatalities as a result of motor vehicle crashes. In
2001, seniors age 85 and over suffered only a tenth of the number of fatalities that teenagers
and young adults ( 20- 24) experienced ( see Figure 2- 6). This fact reflects smaller population in
the elderly as well as reduced driving.
0
1000
2000
3000
4000
5000
6000
7000
15-
19
20-
24
25-
29
30-
34
35-
39
40-
44
45-
49
50-
54
55-
59
60-
64
65-
69
70-
74
75-
79
80-
84
85+
Male
Female
21
Figure 2- 6. Number of Fatal Crashes by Age and Gender, 2001
* For consistency this graph is based 2001 data from the Fatal Analysis Reporting System
( FARS). FARS 2002 is available, but Figure 7 and Figure 8 refer data sources that were most
recently updated in 2001.
Although the absolute number of fatal crashes is lower for the elderly, this does not indicate
that older drivers are safer drivers. After controlling for the number of drivers in each
category, we actually conclude that older drivers have higher fatal crash rates than other
adults. This second data analysis method interests insurance companies and the Departments
of Motor Vehicles because it represents the risk that each driver will be involved in a collision.
Figure 2- 7 shows that the average fatal crash rates per 10,000 licensed drivers is highest for
teenagers, decreases with age until about age 50, and then increases steadily starting at age 65.
0 1 2 3 4 5 6 7 8 9
10
15-
19
20-
24
25-
29
30-
34
35-
39
40-
44
45-
49
50-
54
55-
59
60-
64
65-
69
70-
74
75-
79
80-
84
85+
Male
Female
Both
Figure 2- 7. Number of Fatal Crashes per 10,000 Licensed Drivers, 2001
* Based on the FARS 2001 and the BTS Licensed Driver 2001 database. BTS 2001 is the
latest available national survey of the number of licensed drivers.
Yet another method of analyzing crash involvement is to control for the annual miles driven by
persons in each age category. The number of collisions per mile driven represents “ actual”
risk to the driver, and implies that the more miles he drives, the more likely he will experience
a crash. This method reveals an even starker difference in fatality rates between older adults
and the younger population. Figure 2- 8 shows that adults age 85 and over are involved in
more fatal crashes per mile than any other age group, including teenagers. If this fact remains
true in the coming years, motor vehicle fatalities will be one of the top concerns for elderly
22
drivers and injury specialists as the elderly increase as a percentage of the whole population
and drive more than previous elderly populations.
0
2
4
6
8
10
12
14
16
15-
19
20-
24
25-
29
30-
34
35-
39
40-
44
45-
49
50-
54
55-
59
60-
64
65-
69
70-
74
75-
79
80-
84
85+
Male
Female
Both
Figure 2- 8. Number of Fatal Crashes Per 100 Million Miles Driven, 2001
* Based on the FARS 2001 and the National Household Transportation Survey ( NHTS) 2001.
NHTS 2001 is the latest available national survey on annual miles driven.
Elderly drivers might have very high fatality rates per miles driven, but that does not
necessarily mean that elderly drivers are involved in more forcefully violent crashes than other
drivers. Although poor driver performance may contribute to the fatality rates, older adults
also are far more fragile than younger adults, and are more easily injured and are less likely to
recover from injury than younger adult bodies. Controlling for the mechanical forces in a
crash, older drivers are more likely to die in a crash than younger drivers. 5 Figure 2- 9
illustrates recent driver fragility as a function of age; these data illustrate the fatality rates per
1000 crashes is eight times higher for adults 85+ than for teenagers.
5 Evans, L. Traffic Safety and the Driver. 1991
23
0 5
10
15
20
25
30
35
40
45
50
15- 19
20- 24
25- 29
30- 34
35- 39
40- 44
45- 49
50- 54
55- 59
60- 64
65- 69
70- 74
75- 80
80- 84
85+
Fatalities per 1000 Crashes
Female
Male
Total
Figure 2- 9. Driver Fragility: Fatalities Per 1000 Crashes*
* Based on the California Statewide Integrated Traffic Records System ( SWITRS), all crashes
from 1999 to 2002 ( inclusive).
Figure 2- 9 illustrates a stark difference in fragility rates for males and females. Note that this
difference may be misleading, because this analysis did not control for the physical impact of a
crash. Males are more often cited for speeding violations than females, and therefore may
experience more fatal or serious collisions ( as a percentage of all of their fatal and non- fatal
collisions) than females.
Although seniors are more susceptible to motor vehicle fatalities due to increased fragility,
fragility is not the sole factor for an increase in fatal crashes per mile driven. Figure 2- 10
shows that even non- fatal crash rates per million miles driven increases with age.
0
5
10
15
20
25
15-
19
20-
24
25-
29
30-
34
35-
39
40-
44
45-
49
50-
54
55-
59
60-
64
65-
69
70-
74
75-
79
80-
84
85+
Non- Fatal Crashes Per 1 Million
Miles Driven
Male
Female
Both
24
Figure 2- 10. Non- Fatal Crashes Per Million Miles Driven, 2001*
* Based on the 2001 General Estimates System and the 2001 National Household
Transportation Survey.
To further examine the causes of high motor vehicle injury rates in the elderly, we turn from
fragility and injury rates to specific collision types and traffic violation citations.
Implications for GapAdvise
The increase in both fatal and non- fatal crash risk with increasing age after about the age of
65 means that there will be a very high demand for vehicle and highway design to mitigate this
increasing risk of crashes.
The very sharp increase in fatality ( per crash) with increasing age means that there will be a
very high demand for improved vehicle and occupant restraint design to accommodate and
increasingly fragile population.
2.4 Elderly Drivers and Collision Factors
In order to address the problem of high fatality rates in older drivers, we begin by examining
the kinds of crashes most prominent among older drivers. Collision factors, as well as crash
rates, vary with age. For all drivers, the most common traffic violation attributed to causing a
collision is failure to yield right- of- way. Adults age 70 and over are charged with more than
twice as many right- of- way violations per mile driven than adults age 30 to 60. The second
most common violation for older drivers is failure to obey a traffic signal or stop sign. The
number of traffic violations shown in Figure 2- 11 was obtained from the General Estimates
System, and the rate was computed based on annual miles driven from the National Household
Transportation Survey. Although informative, these data do not detail the primary collision
factors and probably ignore the primary cause of fault of any driver who died in a crash.
25
0
20
40
60
80
100
120
140
160
15-
19
20-
24
25-
29
30-
34
35-
39
40-
44
45-
49
50-
54
55-
59
60-
64
65-
69
70-
74
75-
79
80-
84
85+
Violations per 100 Mil Miles Driven
Alcohol
Speeding
Right- of- Way
Running Signal
Other
Figure 2- 11. Rate of Traffic Violations Contributing to a Crash, 2001*
* Based on the 2001 General Estimates System and the 2001 National Household
Transportation Survey.
A more detailed look at the two most common violations of older drivers, right- of- way and
traffic signal or stop sign collisions, highlights intersection crashes. Older drivers are over-represented
in intersection crashes, and, within these, in crossing path crashes. We analyzed
the following crossing- path crash types6:
1. Left Turn Across Path - Opposite Direction Conflict ( LTAP/ OD)
2. Left Turn Across Path - Lateral Direction Conflict ( LTAP/ LD)
3. Left Turn Into Path - Merge Conflict ( LTIP)
4. Right Turn Into Path - Merge Conflict ( RTIP)
5. Straight Crossing Paths ( SCP)
6 Smith DL, Najm WG. Analysis of Crossing Path Crashes for Intelligent Vehicle Applications. 8th WorldITS
Congress.
26
Figure 2- 12. Intersection Cross- Path Collision Types
Figure 2- 12 shows the distribution of crossing path pre- crash scenarios for driver under and
over age 65. The shaded ( red) vehicle represents the “ subject vehicle” ( usually the turning
vehicle and at- fault vehicle); the other ( white) vehicle represents the “ principle other vehicle”.
First, we will examine the distribution of crossing path pre- crash scenarios ( Figure 2- 13). This
graph shows the percentage of collision- path scenarios for crossing path collisions. For
example, if a driver under 65 is involved in a crossing path collision, it is most likely a straight
crossing path collision because the majority ( 32%) of all crossing path collisions for adults
under 65 is SCP. Similarly, the majority of crossing path collisions for drivers 65 and over is
LTAP/ OD ( 30%). For all drivers, the most common cross- path collision types are SCP,
LTAP/ OD, and LTAP/ LD. For drivers 65 and over, left turns make up 57% of all crossing
path collisions. Of all crossing path collisions, drivers 65+ experience slightly more
27
LTAP/ OD, RTIP, and LTIP collisions than drivers under age 65. ( Other crossing path
collisions – CP OTHER – could be collisions between pedestrians and motor vehicles, a
vehicle making a wrong- way turn onto a one- way street, and other scenarios.)
0
5
10
15
20
25
30
35
Percent
LTAP LD LTAP OD LTIP RTIP RTIP OD SCP CP Other
Under 65
65+
Figure 2- 13. Distribution of Crossing Path Pre- Crash Scenarios, 2002*
* Based on the General Estimates System, data from 2002.
Figure 2- 13 describes the distribution of crossing path pre- crash scenarios, not the actual
number of collisions. Older drivers are actually involved in fewer collisions than younger
drivers. Of all of the collisions in the United States, an older driver is at fault only 7.5% of the
time, and 92.5% of the time, the at- fault driver is less than 65 years old ( from 2002 GES
collision data). However, in the event that an older driver is a collision, they are more likely
to be in a left- turn collision rather than a straight crossing path collision.
The previous paragraph mentions that older drivers make up 7.5% of the at- fault drivers of all
collisions in the United States. However, this percent changes if we look at specific collision
types. For example, of all rear- end collision in the United States in 2002, only 5.7% of the at-fault
drivers were over the age of 65. However, for LTAP- OD collisions, 13.7% of the at- fault
drivers were over the age of 65. These two numbers show that older drivers cause a small
percentage of rear- end collisions, and cause a relatively larger percentage of LTAP/ OD
collisions. These data, as well as other collision types, are summarized in Table 2- 3.
28
(“ OTHER” non- crossing path collisions could be side- swipe collisions, head- on collisions
with fixed objects, and others.)
Table 2- 3. Percent of Drivers 65+, by Pre- Crash Scenario, 2002*
% Over 65
LTAP OD 13.73
LTAP LD 12.25
LTIP 15.12
RTIP 17.82
SCP 10.90
CP
OTHER
9.08
REAR
END
5.73
OTHER 6.02
TOTAL 7.51
* Based on the General Estimates System, data from 2002
Elderly drivers cause a relatively high percentage of crossing- path collisions. This fact is not
the result of elderly drivers driving through more intersections than other drivers. To prove
this, we compare the number of “ subject vehicles” and “ primary other” vehicles at intersection
collisions. For cross- path collisions, drivers age 65 and over more likely to be at fault, or the
“ subject vehicle”, than otherwise (“ principle other vehicle”). Figure 2- 14 shows the
percentage of drivers who are older drivers in different crash types. If all drivers were
involved in the same about of collision type- crashes and were equally likely to be at fault or
not at fault, the graph’s bars would all be the same height. Figure 2- 14 shows that in crossing
path collisions, the at- fault driver is more likely to be older, i. e., over 65 years. This could be
result of different factors. For example, it is possible that older drivers are more likely to drive
locally, and therefore make more turns at intersections than drivers charting a longer distance
and hence more often driving straight through intersections. Regardless of the absence of an
exposure measure, this graph shows a significant difference in at- fault drivers versus
“ principle other” drivers. For all cross- path collision types, drivers over the age of 65 are
more likely to be the subject vehicle rather than the principle other vehicle. Therefore, older
drivers are more often at- fault in cross- path collisions than other drivers.
29
0
5
10
15
20
LTAP
OD
LTAP
LD
LTIP RTIP SCP CP
OTHER
REAR
END
OTHER TOTAL
Percent
SV
POV
Figure 2- 14. Percent Drivers 65+ By Type of Crash and Role in Crash, 2002*
* Based on the General Estimates System, data from 2002.
Not only does the General Estimate Systems data reveal the over- representation of older
drivers as the “ subject vehicle” operator in cross- path collisions, but the data also show that
older drivers are cited for more violations in cross- path collisions ( per mile driven) than adults.
Figure 2- 15 shows the number of violations, resulting from a cross- path collision, cited per 1
billion miles.
0
20
40
60
80
100
120
15- 19
20- 24
25- 29
30- 34
35- 39
40- 44
45- 49
50- 54
55- 59
60- 64
65- 69
70- 74
75- 80
80- 84
85+
CP Violations per Bil Mile Driven
Right of Way
Running Signal
Figure 2- 15. Rate of Violations in Cross- Path Collisions, 2001*
* Based on the 2001 General Estimates System and the 2001 National Household
30
Transportation Survey.
Implications for GapAdvise
The increase in both right- of- way violations and running- a- stop- signal violations will lead to
a high demand to increase driver on- road and intersection awareness through vehicular and
roadway instrumentation.
As older drivers are over- represented in intersection collisions, there will also be a high
demand to introduce instrumentation that augments the driver’s ability to make safe decisions
about when to enter an intersection.
2.5 Causal Factors
Understanding and describing driver behavior becomes a challenge when one tries to identify
driver errors in determining crash causal factors and countermeasures. Access to data related
to crashes is usually based on crash statistics and restricted to general characteristics of the
involved drivers, such as gender, age, type of vehicle driven. Very rarely are the actions and
maneuvers that led to a crash addressed. This section briefly highlights some previous research
that focuses on the causal factors of older drivers’ crash rates.
The investigation of pre- crash actions and maneuvers usually relies on either focus groups
involving officers who respond to crashes or drivers involved in crashes. 7 They therefore rely
on subjective sources. Another approach adopted for understanding why crashes occur consists
of linking general characteristics with known issues of specific group, such as age linked with
perceptive and cognitive deficits. 8
Staplin and Fisk investigated older drivers’ difficulties with intersections. 9 The underlying
7 Wierville W. W. Hanowski R. J. Hankey J. M Kieliszewski C. A. Lee S. E., Medina A. Keisler A. S and Dingus
T. A. ( 2002) Identification and evaluation of driver errors: overview and recommendations FHWA- RD- 02- 003.
8 Hakamies- Blomqvist, L. ( 1996) Research on older drivers: a review. IATSS, 20( 1), pp. 91- 101.
9 Staplin L., Fisk A. D., ( 1991) A cognitive engineering approach to improving signalized left turn intersections
Human Factors 33 ( 5) 559- 571
31
causes were identified to be perceptive and cognitive problems. “ Perceptive” can be defined in
terms of visual acuity and contrast sensitivity lost. “ Cognitive” relates to working memory and
information processing. Also, the assumption that presenting information in advance would aid
older drivers was not shown true, as this did not help older drivers to make a faster decision in
the end.
The importance of both perception and cognition in driving tasks arises in other studies as
well. Larsen and Kines reported on an extensive investigation of crashes in Denmark. 10 The
main problems they identified for left turning drivers are attention errors and misjudgment of
the time available to complete the maneuver. None of the cases they investigated was due to a
driver who misunderstood the right of way.
Hancock and Caird focused on the assessment of the appropriate time to turn left with variable
oncoming traffic speed and time gap size. 11 They concluded that decisions do not depend only
on velocity or gap size but on some cue extrinsic to these parameters. Older drivers seem to be
more conservative than young. Both young and old drivers do not initiate turns upon oncoming
velocities, gap size or distance; rather, they use higher order information extracted from these
parameters, like time to arrival or rate of frontal expansion.
Implications for GapAdvise
Focus groups, observational studies, and driving experiments ( as used by other researchers)
are the best means of measuring driver decision making and behavior at intersections.
Future instrumentation to augment drivers’ decisions at intersections should address attention
errors and gap misjudgment.
10 Larsen L. and Kines P. ( 2002) Multidisciplinary in- depth investigations of head- on and left- turn road collisions
in Accident Analysis and Prevention ( 34) 367- 380
11 Caird J. K. and Hancock P. A. ( 2002) Chapter 19: Left turn and gap acceptance crashes in R. E. Dewar & P.
Olson ( Eds) Human factors in Traffic Safety 736 p
32
3.0 CONDUCT FOCUS GROUP AND OBSERVATIONAL ANALYSIS OF ELDERLY
DRIVERS
3.1 Focus Group Research
Safe older driving was explored in four focus groups conducted in July, August, and
September of 2004 at the Rossmoor Senior Adult Community in Walnut Creek, California ( see
Appendix A for detailed summaries of each focus group, the focus group protocol). The 20
women and 16 men who participated in the focus group were Rossmoor residents who drove,
were between the ages of 70 and 85, and passed a screening test of physical and cognitive
acuity ( see Appendix A). This summary describes the general findings from all four focus
groups.
3.1.1 Demographic and Attitudinal Profiles
At the beginning of each focus group a questionnaire was administered that explored the
demographic attributes of focus group participants, their travel patterns, and their attitudes
toward various transportation modes ( see Appendix A). The results for all participants in the
four focus groups are examined here.
In Table 3- 1, below, data on vehicle type by gender and age are presented. Participants drove a
range of vehicles, from small compacts to luxury sedans, manufactured by a variety of
automakers.
Table 3- 1. Vehicle Type by Age, Gender and Focus Group
Focus Group Gender Age ( Years) Car Make/ Model
1 F 72 Hyundai Elantra
1 F 74 2001 Hyundai Elantra
1 F 78 1994 Toyota Tercel
1
F 79
Do not drive household car – don’t know
make/ model of vehicle
1 F 83 1995 Buick Century
1 F 83 1998 Chevy Malibu
1 M 71 1996 Dodge Intrepid
1 M 73 1998 Toyota Corrolla
1 M 78 1993 Dodge Shadow
1 M 81 1994 Mercedes E420
2 F 76 2000 Dodge Durango
33
2 F 76 2001 BMW 325i
2 F 77 1996 Toyota Camry
2 M 70 2002 Toyota Camry
2 M 77 2001 Ford VXZ Escort
2 M 78 2004 Lexus RX330
2 M 81 2000 Dodge Caravan
2 M 83 1993 Lexus ES300
3 F 71 1998 Toyota Camry XLE
3 F 75 1995 Saturn Wagon
3
F 76
Do not drive household car – don’t know
make/ model of vehicle
3 F 81 2003 Toyota Corolla
3 F 84 1998 Honda Accord LX
3 F 85 2002 Honda Accord
3 M 73 1999 Acura Integra
3 M 82 2000 Lexus 300 ES
3 M 83 1991 Toyota Corolla
3 M 84 1996 Toyota Camry
4 F 74 1998 Lexus sedan
4
F 79
Do not drive household car– 2004 Honda
Civic
4 F 80 2004 Hyundai Sonata
4 F 81 2002 Mercedes C240
4 F 83 1988 Toyota Camry
4 M 74 Buick LeSabre
4 M 74 1996 Volvo 850
4 M 79 1996 Mercury Sable
M= male and F= Female
Note: Kathryn Hamel, Ph. D., provided the data in this table.
Aggregate demographic attributes of all participants in the four focus groups are provided in
Table 3- 2 ( below). The average focus group participant:
· Was 78 years old and married;
· Had a Bachelor’s degree and an income between $ 20,000 to $ 49,000;
· Lived in a household with 1.5 people, 1.5 drivers, and 1.4 autos; and
· Had been driving since s/ he was 18.5 years old.
34
Table 3- 2. Demographic Attributes
Mean ( N= 36)
Age 78
Household Size 1.5
Household Drivers 1.5
Household Autos 1.4
License Age 18.5
Distribution
Income
< $ 10,000 6%
$ 10,000-$ 19,000 3%
$ 20,000-$ 49,000 33%
$ 50,000-$ 79,000 14%
>$ 110,000 14%
Declined to Respond 31%
Marital Status
Single 8%
Married 58%
Divorced 8%
Widowed 25%
Education
High School 9%
Associate's Degree 17%
Bachelor's Degree 50%
Graduate Degree 14%
The travel modes used more than two times per week by focus group participants are presented
in Table 3- 3 ( below). Driving alone was the most frequent travel mode, followed by walking,
and the Bay Area Rapid Transit ( BART) District transit system.
Table 3- 3. Frequently Used Travel Modes
Percentage
Drive Alone 97%
Carpool 6%
Bus 6%
BART 11%
Walk 47%
Note that the total sums to more than 100 percent because respondents indicate use of more
35
than one mode.
The types of services and devices used by focus group participant are presented in Table 3- 4
( below). Most participants used both cellular phones and the Internet.
Table 3- 4. Devices and Services Used by Participants
Percentage
Cellular Phone 3.1%
Internet 25.0%
Both 71.9%
The survey instrument also explored participants’ travel- related attitudes, with results shown
in Table 3- 5. Questions examined participants’ perception of vehicle hassle, experimentation,
vehicle enjoyment, and overall vehicle satisfaction. Vehicle enjoyment is a different criterion
than vehicle satisfaction; many participants claimed to enjoy driving as a recreational activity
( enjoyment), others, to be satisfied with it as a means of mobility ( satisfaction). The focus
group participants generally agreed or strongly agreed that they enjoyed and were satisfied
with their vehicle. In addition, they were generally neutral towards vehicle hassle ( e. g., costs
and frustrations associated with vehicle ownership and maintenance, including taking cars in
for repairs and finding parking) and experimentation ( i. e., attitudes towards trying new things,
such as advanced technologies).
Table 3- 5. Attitudinal Factors
Factor Score
Vehicle Hassle 3.2
Experimentation 3.3
Vehicle Enjoyment 4.1
Satisfaction 4.5
Scale: 1= strongly disagree, 2= disagree, 3= neutral,
4= agree, 5= agree strongly
Finally, the survey explored the frequency with which the participants used transit, currently
and in the past, as well as barriers to driving and transit use that may have influenced their
choices. These results are summarized in Table 3- 6 ( below). For some questions, the sample
size was smaller because less than half ( 15) of the participants used transit more than
36
occasionally. The results indicate that:
· In the past, 42 percent of participants regularly used transit at some time in their life
before moving to Rossmoor;
· Currently, only 14 percent always or usually use transit, but 31 percent sometimes use
the service;
· Few participants indicated difficulties with physical barriers to transit use ( e. g., stairs,
stepping off the bus, and purchasing tickets);
· Sixty percent or more of the participants sometimes chose to take transit when the
alternative was to drive in bad weather, heavy traffic, or unfamiliar areas; and
· Insensitivity to transit cost and travel time was expressed by many participants
Table 3- 6: Factors Influencing Frequency of Transit Use
N= 36
Previous Regular Transit Use 42%
Current Frequency of Transit Use N= 36
Never/ rarely 53%
Sometimes 31%
Always/ usually 14%
Physical Barriers N= 15
Stepping Off Bus or Train 7%
Station Stairs 13%
Purchasing Tickets/ Paying Fee 7%
Take Transit ( At Least Sometimes) To Avoid… N= 15
Driving at Night 20%
Left Turns 47%
Bad Weather 67%
High Traffic Roads 60%
Unfamiliar Areas 60%
Avoid Transit ( At Least Sometimes) If It… N= 15
Costs More 27%
Takes Longer 40%
New Schedule 33%
Transfer 33%
Involves a New Transit Station or Stop 20%
Note: N= 15 excludes participants who never or rarely use transit.
37
3.1.2 Synthesis of Focus Group Discussions
Introductory Comments on General Travel
Although participants in all four groups were aware of their limitations as older drivers, they
expressed an overwhelming preference for travel by automobile and most used transit
infrequently. Some were concerned about driving at night and during bad weather, but most
had little difficulty with congestion. Residents reported very little difference between their
travel behavior on weekends and weekdays despite heavier weekday traffic. Congestion was
cited, however, as a reason for using the Bay Area Rapid Transit ( BART) system for travel to
San Francisco, and some participants avoided peak- hour traffic. Overall, however, their
mobility was not limited by adverse driving conditions.
Accessing Vehicles
During the focus group discussions, participants discussed different aspects of getting in and
out of their car, including their use of remote keyless entry, difficulties loading packages into
the trunk or back seat, and their use of seat adjustments ( both manual and automatic).
Remote Keyless Entry. More than half of the participants had keyless entry devices for their
automobiles, and those who did described a variety of benefits of their use, including locating
their parked vehicles and locking/ unlocking their car doors when unloading or loading
packages. Feelings about the alarm feature installed with the device were mixed, and some
residents reported disarming the feature because it was too easy to activate accidentally. Some
had malfunctioning devices or had difficulty learning how to use them correctly, but it
appeared that once residents became familiar with their use these concerns were outweighed
by the technology advantages.
Loading and Unloading Vehicles. Several participants noted the advantage of using the trunk
over the back seat for transporting packages ( i. e., additional privacy and more space for large
items). However, nearly all used the floor or back seat at least occasionally because they felt
that items in the trunk were likely to slide out of place during driving. Suggestions made for
38
resolving this problem focused on low- technology, cost- effective solutions that many residents
had already installed in their vehicles, including netting, bungee cords, foam mattresses, and
removable partitions.
Others found that high trunk lips in the back of their vehicles made lifting heavy items into the
trunk difficult. Although sport utility vehicles and station wagons already have flat trunks that
make loading easier, most residents drove sedans or other automobiles without this feature.
Seats and Seat Adjustments. Taller residents and those with disabilities often had difficulty
getting in and out of their cars, and all felt that adjustable seats made the maneuver easier. Seat
adjustment, however, did pose some additional difficulties. Residents, particularly those who
shared their cars with a spouse or partner, disliked having to move the seat back after it had
been adjusted by another driver. There was an overwhelming preference for cars with preset
adjustments for multiple users.
Petite residents, who often had to raise their seats to see over the dashboard, were concerned
about being too close to the steering wheel during a crash and felt that airbags should be
redesigned for safe deployment.
Seat type was another concern for many drivers. Some had difficulty getting into cars with low
bucket seats, and others had difficulty adjusting them. Overall, however, there was no
consensus about which seat type was most comfortable.
Other Concerns and Recommendations. Doors were also a concern for several participants,
who felt that they often did not open widely enough. Others complained about doors that
closed unintentionally while they were getting in or loading packages; several suggested that
door stops would make access easier. One resident drove a car with an adjustable steering
wheel and found that this helped with getting in and out of the vehicle.
39
Driving
Focus group participants described a variety of difficulties operating their vehicles. For older
drivers, neck turning can often be physically difficult, and many residents expressed concern
with blind spots and gauging distances in their side- view mirrors. As a result, the primary
problems the drivers experienced were with difficult maneuvers that require a broader field of
vision, including parallel parking, reversing, merging, and making left- hand turns. Night
driving was also mentioned as problematic.
Parking and Reversing. Several residents expressed frustration with parallel parking. In
particular, most had difficulty seeing behind them while reversing because of blind spots, and
there was general agreement that " wink" mirrors, which provide a broader field of vision, were
preferable. Other car enhancements that were viewed favorably included a remote- adjustable
rear- view mirror and a global positioning system ( GPS)- enabled camera that allows drivers to
see behind them while fitting into a tight space.
In general, participants felt that reversing was dangerous and suggested that their vehicles be
equipped with beepers or other devices to signal their presence to pedestrians or other vehicles.
Merging and Left- hand Turns. Although it was initially assumed that drivers would be
principally concerned with making difficult left- hand turns, focus group participants instead
expressed a much greater concern for both merging onto the freeway and changing lanes. In
both cases, however, the causes of this concern were the same: difficulty gauging distances of
oncoming traffic using convex mirrors and trouble seeing other cars because of blind
spots ¾ particularly prevalent among drivers who had difficulty turning their necks. In
particular, pillars in the back seat windows were identified as obstructions to the view behind
the vehicle. Several participants felt that other drivers were reluctant to slow down at high-speed
merge points.
Some drivers went out of their way to avoid left- hand turns, citing a similar set of concerns
and a lack of left- hand turn lanes in certain localities. Because of this sense of perceived
40
control ( i. e., it is possible to make three right turns to avoid making a left one) left- hand turns
were not identified as being as serious a difficulty as merging into a right hand lane or as other
maneuvers, which are often unavoidable.
Participants also noted that many drivers leave their turn signals on longer than necessary and
suggested that manufacturers install devices that automatically shut them off after a specified
period of time or make the audio alerts louder for the hearing- impaired.
Night- Time Driving. Several drivers complained about glare from incoming headlights and
inquired whether cars could be equipped with automatic dimmers to lessen this problem.
Another driver spoke highly of a vehicle he had once driven that had headlights that pivoted
with the wheels, improving visibility while turning. The specific vehicle model was not
identified, however.
Vehicle Use
Residents also had difficulty with features on their cars that were not directly related to
driving, including display panels, knobs, and dials. Participants also expressed their opinions
on the use of navigation aids and cell phones.
Dashboard Displays. Participants noted a variety of difficulties with their dashboard displays.
Some had trouble reading the LED displays because they were not bright enough or too similar
to the background panel color. One resident was unable to read the digital clock in his vehicle
because of the angle of the dashboard. Another complained that the steering wheel obstructed
his view of the dashboard. In general, participants expressed support for digital compasses
mounted in their dashboards.
Radios and Radio Adjustments. Several participants had difficulty adjusting their radios and
rarely used them or only used them in light traffic. Suggestions included using push buttons
rather than more- difficult- to- operate knobs, which could assist with dexterity difficulties and
provide pre- set access to favorite radio stations. Participants also thought that installing
41
controls near the steering wheel for easier access, and providing remote controls might be
helpful, but they did not have direct experience with these features.
Cell Phones. Most of the participants had cell phones but few admitted to using them while
driving. When asked, there was widespread support for laws against in- vehicle use of mobile
phones.
Maps and Guides. Residents were often familiar with the online service MapQuest  , but
several found that the routes provided were occasionally circuitous. Several participants only
used traditional paper maps. In general, participants were reluctant to identify cognitive
difficulty with receiving directions or reading maps, but they were enthusiastic about readily
accessible, in- vehicle information such as GPS or on- board compasses.
In- Vehicle Navigation. Those participants who had in- vehicle navigation systems spoke
favorably of them. Some were concerned about the distraction of a GPS screen, but the
primary concern of most residents was system cost.
3.1.3 Study Limitations
The focus group research methodology allows for detailed, in- depth exploration of relatively
new research areas, but its small, non- random sample limits generalizations to the larger
population. As a result, it is important to interpret the results of the focus group findings in the
context of the demographic and attitudinal profiles of the participants, as described in detail
above. More specifically, the sample was drawn from residents of the Rossmoor Senior Adult
Community ( Walnut Creek, CA) who are, on average, wealthier than members of a random
sample of older drivers drawn from the larger population. In addition, participants were
screened for physical and cognitive acuity ( a requirement of the University of California
Human Subjects Review of the study - see Appendix A). Thus, participants in this study do not
represent the frailest or most impaired drivers.
Researchers also made two observations about the hesitation among participants to discuss
42
their driving impediments. The first was that male participants were often less forthcoming
with their physical and cognitive challenges than were females. The second was that
participants appeared less willing to talk about cognitive difficulties with driving ( e. g., getting
lost or merging/ turning decisions) than they were about physical ones ( e. g., difficulty turning
their necks). Because cognitive challenges are more difficult to observe in biometric tests, the
relationship between cognitive disability and safe driving should be studied in more detail than
was possible here.
3.2 Observational Research
The link between specific impairments and “ in- vehicle” performance has been previously
investigated using laboratory settings, instrumented cars, and closed- road circuits, which
involve driving a set course without other vehicles present. Additionally, these studies have
primarily used in- vehicle testers to assess impairments and infractions. Past studies have not
established an association between functional assessment tests and “ in- vehicle” performance
in an open- road scenario using the subject’s own vehicle.
Porter and Whitton ( 2002) 12 established the use of the Global Positioning System ( GPS) and
“ in- vehicle” video technology to detect age- related differences during driving performance in
the subject’s own vehicle. This system allows the driver to perform in a less imposing test
environment in comparison to other methods used in the past. Porter and Whitton also
recorded the driving scene with video technology, but did not record the driver during
performance. While the analysis of the driving performance can be blinded with this set- up,
crucial knowledge of the driver’s physical activity is lost. Studies that analyze the interaction
between the driver’s abilities and the driver’s performance within his/ her own vehicle, provide
crucial information to the public and the motor vehicle industry.
Once specific impairments of older adults are factored into the equation to predict “ in- vehicle”
performance, research regarding possible intervention strategies can be addressed. Physical,
cognitive, and visual medical intervention, as well as motor vehicle modifications could be
12 Porter, M. M. and M. J. Whitton, Assessment of driving with the global positioning system and video
technology in young, middle- aged, and older drivers. J Gerontol A Biol Sci Med Sci, 2002. 57( 9): p.
43
used to address the problem of elderly driver safety. Research indicates there is a need to
explore modifications of private vehicles and the use of technology to enhance the
performance of older drivers13. Use of GPS and video technology, combined with assessment
of the driver, vehicle, and the driver’s concerns regarding their vehicle, could lead to a safer
driving experience on all roads.
The specific aims of this subtask were to observe and analyze older adults during “ in- vehicle”
performance on an open road course and also during ingress/ egress tasks. Additionally, we
sought to document the effectiveness of GPS and video technology to assess “ in- vehicle”
performance of older drivers. It was hypothesized that problems faced by older drivers would
be clearly observed through analysis of “ in- vehicle” performance. It was also hypothesized
that the problems detected in this study would direct future research on specific intervention
strategies to address these problems. Future motor vehicle modifications, along with medical
and behavioral intervention strategies should be targeted at keeping older drivers safe on the
road, despite functional declines.
3.2.1 Methods
Subject Population
Sixteen men ( average age = 77 ± 5 yrs; range = 70- 84 yrs) and twenty women ( average age =
78 ± 4 yrs; range = 71- 85 yrs) were recruited to take part in an observational video analysis of
vehicle use and a focus group ( reported in Section 3.1.1) on extending safe driving years for
older adults. The study received Institutional Review Board approval through UCSF and UC
Berkeley. Subjects were recruited from the Rossmoor community in Walnut Creek, CA,
which consisted of 6,700 residential units, including co- operatives, condominiums, and single-family
home developments. In order to reside in Rossmoor, one resident per dwelling must be
at least 55 years of age and all residents must be able to live independently. Further
information on the Rossmoor community can be found at www. Rossmoor. com. Subjects were
13 Shaheen, S., Niemeier, DA, Integrating vehicle design and human factors: minimizing
elderly driving constraints. Transportation part C, 2001. 9: p. 155- 174.
44
recruited through flyers posted throughout common areas in Rossmoor and an article in the
Rossmoor News. Exclusion criteria for the study included:
1. Having a history of neurological disease likely to affect neuromuscular function
including a stroke ( Cerebral Vascular Accident), seizure disorder, or Parkinson’s.
2. Having a diagnosis of dementia or Mini- Mental Status Examination score < 24.
3. Standard visual acuity worse than 20/ 40.
4. Having a history of any other previous illness or surgery, such as a vestibular disorder,
significant visual disorder, arthritis, or cardiovascular disease, which might, in the
opinion of the investigator, interfere with normal driving behavior.
5. Currently taking any medications that might interfere with driving.
6. Did not currently hold a valid California driver’s license.
7. Did not currently drive at least 3 days per week.
8. Did not own/ lease their own vehicle.
9. Had been involved in a motor vehicle accident or DUI within the last 2 years.
10. California car license and registration were not valid and current
11. Proof of liability insurance did not meet the minimum liability requirements of $ 50,000
for death or injury of any one person, any one accident; $ 100,000 for all persons in any
one accident; and $ 25,000 property damage for any one accident ( California DMV
registration requirements are $ 15,000/$ 30,000/$ 5,000).
Specific Procedures
Pre- screening, included the Telephone Interview for Cognitive Status ( TICS), which is similar
in content to the Mini- Mental Status Examination. Questionnaires on general health, driving
activity, and driving confidence were sent out to subjects ( Appendix B), completed at home,
and subsequently brought in by each subject on the day of testing. All participants voluntarily
consented to take part in the study. Participants reviewed and signed a consent form
acknowledging awareness of the study purpose and risks associated with participation.
Subjects were paid $ 25 for the driving session as compensation for costs of vehicle use and
time and received an additional $ 75 after participating in the focus group.
45
Intake Examination
After completing the informed consent process, physical, visual and cognitive function of each
participant was assessed with a 2- hour battery of measurements listed Table 3- 7. The subject
was required to complete the intake tests before participating in the driving portion of the
study. If information attained from the medical history questionnaire or intake assessment led
the investigators to think a condition or impairment could interfere with normal driving, the
subject was not allowed to perform the on road portion of the testing. If excluded, the
participant was still allowed to take part in the focus group.
Package Loading and Ingress/ Egress
After completion of intake examination measures, subjects were asked to perform the task of
putting a bag of groceries and suitcase into their vehicle “ as they normally would.” Each item
weighed 10 pounds. Subjects were videotaped during the loading of packages and during
ingress and egress from the driver’s seat and the rear passenger seat ( rear passenger seat
evaluation was added to the test battery after the first 9 subjects had completed the study).
Package loading was evaluated from the video and scored on placement ( backseat, floor of
backseat and trunk) and difficulty. Ingress and egress were evaluated for difficulty compared
to a young healthy adult performing the same tasks ( see scoring criteria in Appendix B).
46
Table 3- 7: Intake Examination
Physical Range of Motion Instrument
Cervical Spine Active Range of Motion
( AROM): Rotation
CROM: head mounted goniometer
Gross Upper Body AROM Driving Health Inventory:
Head- neck- thoracic spine rotation test: requires the participant to turn their
whole body to see an object on a computer screen 10 feet behind their
chair
Lower Extremity AROM:
Ankle, knee and hip motion
Hand- held goniometer: available motion at the ankle, knee and hip was
assessed as the participant actively flexed or extended each joint
Vision Instrument
Visual scanning PC- Based version of the Trails A and B tests ( Driving Health Inventory):
Asks participant to connect numbers, or letters and numbers, in a
sequential order while they are being timed
Visual closure Motor- Free Visual Perception Test ( Visual Closure subtest; Driving Health
Inventory): Asks participants to determine which “ unfinished” figure
accurately resembles the “ finished” figure
High and low contrast acuity Scan Chart test ( Driving Health Inventory): examined the participants
visual acuity during high and low contrast conditions and at levels of 20/ 40
and 20/ 80
Stereoscopic vision ( Depth Perception) Frisby Stereopsis Test: Asks participants to determine which of four
figures has a “ circle in depth” on a series of plastic cards and a different
viewing distances
Divided attention; Visual processing UFOV- Useful Field of View: The area from which one can extract visual
information in a brief glance without head or eye movement. The limits of
this area are reduced by poor vision, difficulty dividing attention and/ or
ignoring distraction, and slower processing ability.
Low contrast vision Pelli- Robson Contrast Sensitivity: Participants are asked to identify letters
at decreasing levels of contrast
Strength
Instrument
Grip Hand held dynamometer ( force measuring device)
Plantarflexion ( calf muscle) Repeated single leg toe raises up to 25 on each leg
Dorsiflexion ( ankle muscle), Knee Extension
( Quadriceps – thigh muscle)
Hand- held dynamometer
Sit- to- Stand Time Time it took each participant to complete 5 sit- to- stand- to- sit trials as fast
as they could ( could not use their hands and arms to help)
Balance Instrument
Longest time the participant could stand on one leg
Cognition Instrument
Working Memory Delayed Recall test ( Driving Health Inventory): Asked participant to
remember and recall three words at a latter point during testing
47
Driving Performance
Following the assessment of package loading and ingress/ egress, subjects were asked to drive
a pre- determined loop within Rossmoor followed by an approximately 5- mile course to
downtown Walnut Creek, CA, “ as they normally would” ( see Figure 3- 1). The course, which
began and ended at the Rossmoor clubhouse parking lot, allowed the subject to choose their
route to and from the downtown area once they left the Rossmoor gates ( and after following
the prescribed route inside of Rossmoor). Subjects were asked to park in any downtown
parking space ( within a pre- defined area shown to them on a map) and promptly return.
Subjects were told to return without parking if they are unable to locate a space within 10
minutes. The driving course began with subjects backing out of a parking space and included
numerous turns and lane changes. Driving performance for the Rossmoor course and the
section from the gates to downtown Walnut Creek was analyzed for infractions based on a
more detailed modification of the California Department of Motor Vehicles Road Test
( scoring criteria located in Appendix B).
48
= Rossmoor clubhouse = Downtown Walnut Creek, CA
Figure 3- 1. Rossmoor driving route and location of downtown Walnut Creek, CA
A global positioning system was temporarily mounted to the vehicle to monitor driving speed
and location of the vehicle. Additionally, investigators utilized a four- camera “ surveillance”
system integrated with a computer to monitor each subject’s automobile use before, during,
and after completing the driving course ( see Figure 3- 2). The cameras were attached to the
subject’s own vehicle using various clamps and suction cups. The subject drove alone in the
vehicle without anyone else present. Video data were analyzed at a later time for driving
infractions and physical movements of the driver. Infractions were judged simultaneously by
two investigators using scoring criteria developed for use in the study ( Appendix B).
49
Figure 3- 2. Camera views during driving assessment
3.2.2 Equipment
A mobile digital video recorder ( Model 5308; March Networks, Ottawa, Canada) was used to
collect video data from 4 cameras and position and speed data from WAAS- enabled
differential GPS ( Model NCT- 2030M; Navcom Technologies). The video data were sampled
at 15 Hz per camera. For the first half of the study, we attempted to collect GPS data sampled
at 5 Hz, with a positional accuracy of 0.5 m. The position and speed of the vehicle were
automatically integrated and synchronized with the video data and “ time stamped” on the
video output. Unfortunately, due to the location of Rossmoor in Walnut Creek, CA ( within a
valley) and the position of the available satellites over Walnut Creek during the daytime hours
of the summer of 2004, we were unable to collect accurate and reliable GPS data. Therefore,
we used the camera that originally faced the rear of the vehicle ( see Figure 3- 2) and we
50
mounted it to clearly view the speedometer so that we could collect information on the speed
of the vehicle.
3.2.3 Data Analysis
This is a descriptive, observational and correlational study of a group of older adult subjects.
Descriptive statistics were used to describe the performance of the participants.
A correlation matrix was created to examine the relationships between dependent ( intake
examinations measures) and independent ( driving performance) variables. The Spearman rank
correlation coefficient was used for all comparisons.
3.2.4 Description of Participants
The participants in this study were on average, 78 years old and 64% were retired at the time
of the study. Most participants drove at least 6 out of 7 days of the week and typically drove
about 120 miles during the course of one week. Additional demographics can be found in
Table 3- 8 and in the Focus Group Report.
Table 3- 8. Demographics of participants
Mean ± Standard
Deviation
Average Number of Days Per Week Driven 5.9 ± 1.6 days
Average Number of Miles Per Week Driven 118 ± 71 miles
Number of Years the Participants Had Been Driving 58.8 ± 7.8 years
Number of Years the Participants Had Lived in Rossmoor 8.7 ± 6.5 years
Number of Years the Participants Had Lived in Walnut Creek,
CA
17.3 ± 15.4 years
The percentage of participants reporting specific health conditions can be found in Figure 3- 3.
Of particular note was the percentage of participants with arthritic conditions such as
osteoarthritis and rheumatoid arthritis ( 40%) which might limit their ability to get into and out
of a car and manipulate controls in the vehicle. Nearly all participants wore some type of
51
glasses ( 35/ 36) and 45% required the use of hearing aids. Although as whole, the participants
in this study were a relatively robust and high functioning group, they still presented with
many typical age- related disorders and diseases.
0
10
20
30
40
50
60
Heart Attack
Transient Ischemic Attack
Angina
High Blood Pressure
Stroke
Diabetes
Neuropathy
Respiratory Disorders
Multiple Sclerosis
Other Neurological Disorders
Osteoporosis
Rheumatoid Arthritis
Other Arthritis
Vision Problems
Inner Ear Problems
Other Movement Disorders
Depression
Cancer
Joint Replacement
Cognitive Disorder
Uncorrected Visual Problems
Percent of participants reporting problem
Figure 3- 3. Health status among participants
Driving Confidence and Avoidance Questionnaires
A surprising number of participants reported some level of driving avoidance behavior ( see
Figure 3- 4). Driving at night, in bad weather and in unfamiliar situations were the situations
that the participants reported avoiding most frequently. Results included:
· 43% of participants reported that they sometimes or usually avoid driving at night
· 28% of participants reported that they sometimes or usually avoid making left turns
across traffic
· 38% of participants reported that they sometimes or usually avoid driving in bad
weather
52
· 25% of participants reported that they sometimes, usually or always avoid driving on
high traffic roads
· 39% of participants reported that they sometimes, usually or always avoid driving in
unfamiliar areas
· 14% of participants reported that they sometimes or usually pass up opportunities
because of concerns about driving
Do You Avoid...?
0
10
20
30
40
50
60
70
80
Driving at Night Making Left Turns
Across Traffic
Driving In Bad
Weather
High Traffic
Roads
Driving in
Unfamiliar Areas
Opportunities
Because of
Concerns About
Driving
Percent of Participants
Never
Rarely
Sometimes
Usually
Always
Figure 3- 4. Participant responses when asked which type of driving activities they avoid
and how often they avoid them.
Women reported less confidence in their driving ability when compared to the men ( Average
for men = 93/ 100 ( Range = 81- 100); Average for women = 75/ 100 ( Range = 37- 100); t- test: p
= 0.001) ( See Table 3- 9).
53
Table 3- 9. Participants confidence levels in their ability to perform certain driving tasks
Driving Task
Median Score Score Range
Driving at night Men = 10
Women = 8
Men = 3- 10
Women = 2- 10
Driving in bad weather Men = 9
Women = 7.5
Men = 5- 10
Women = 3- 10
Driving in rush hour or heavy traffic Men = 10
Women = 8.5
Men = 7- 10
Women = 1- 10
Highway driving Men = 10
Women = 9
Men = 9- 10
Women = 2- 10
Driving during long trips Men = 10
Women = 9
Men = 9- 10
Women = 0- 10
Changing lanes on busy streets Men = 10
Women = 8
Men = 7- 10
Women = 4- 10
Reacting quickly Men = 9
Women = 8.5
Men = 6- 10
Women = 3- 10
Pulling into traffic from a stop Men = 10
Women = 8
Men = 7- 10
Women = 5- 10
Making a left turn across traffic Men = 10
Women = 9
Men = 7- 10
Women = 4- 10
Parallel parking or backing into space
between cars
Men = 10
Women = 8
Men = 7- 10
Women = 2- 10
Physical/ Musculoskeletal Function Status
The older adults in this study had decreased range of motion in their hips, knees, neck and
spine when compared to the norms for younger adults:
· Participants had approximately 10 ° less hip flexion compared to adults under the age of
60
· Participants had approximately 16 ° less knee flexion compared to adults under the age
of 60
· 50% of participants were unable to turn far enough to see directly behind themselves
while sitting
o limited cervical and thoracic spine range of motion
Women had only 40% of the grip strength of men ( Men = 32 ± 6 lbs; Women = 13 ± 5 lbs)
and 60% of the thigh ( quadriceps) muscle strength of men ( Men = 62.5 ± 15 lbs; Women =
54
39.5 ± 9 lbs).
Cognitive Status
All participants were screened for cognitive ability before the start of the study and scored at
least 4 points above the minimum cut- off level on the cognitive screening test ( TICS).
Although the participants all passed the screening exam, 30% of participants had mild deficits
in working memory and 30% had serious deficits in working memory.
Visual Function Status
A surprising number of participants had deficits in divided and selective attention and directed
visual search tasks. Additional deficits in low contrast visual acuity and visualization of
missing information were also noted:
· 27% of participants had mild or serious deficits in low contrast visual acuity
· 22% had mild deficits and 14% has serious deficits in visualization of missing
information
· 78% had mild deficits and 6% had serious deficits in directed visual search
· 51% had difficulty with divided attention on the Useful Field of View Test
· 32% had difficulty with selective attention on the Useful Field of View Test
Driving Performance
The driving performance of 30 out of 36 participants was evaluated for this report. Three
participants were not allowed to drive in the study based on very low intake scores or other
disqualifying criteria. Three other participants could not be scored because of equipment
malfunction that impaired the ability of the investigators to accurately score driving
performance. Make, model and year of each participant vehicle are listed in Table 3- 1. The
results of the remaining 30 participants are described in detail below.
55
3.4.5 Results: Rossmoor
Overall, participants made more frequent errors in the Rossmoor section of the course than
they did once outside the gates of Rossmoor. This was a surprising finding considering that
most participants complained about “ other drivers” in Rossmoor, but seemed unaware of their
own poor driving behavior. Five participants made critical errors during the Rossmoor section
of the course. These errors, if made during a DMV examination, would have constituted a
failed road test and immediate termination of the exam. All critical errors occurred at three- or
four- way stops. The errors included failing to stop, failing to yield right of way and driving
straight through an intersection from a turning lane.
Four participants did not follow the prescribed route in Rossmoor. When examining the four
participant’s working memory scores and cognitive screening test scores, nothing of note stood
out from other participants.
Of particular note in both Rossmoor and in Walnut Creek were head turning errors. Upon
leaving the staring parking space in the clubhouse parking lot, the majority of participants did
not fully scan behind their car before backing out. The turn out of the Rossmoor parking lot
onto the road is uncontrolled and there were often numerous pedestrians in the immediate
vicinity. However, most participants made at least one error leaving the parking lot.
Key results from the Rossmoor section include:
· 75% of those drivers who reversed out of the starting parking space did not fully look
through rear window before backing out
· 100% of those who pulled forward out of the parking space made no scanning errors
· Many errors were made during turn out of Rossmoor parking lot:
o 90% did not fully stop before turning
o 43% did not scan the surrounding area adequately
56
o 20% failed to slow
o 23% failed to signal
· 40% of drivers made head turning errors at stop sign controlled intersections
· 67% of drivers made head turning errors during lane changes
· 17% of drivers made head turning errors during yield
· 13% of drivers made signaling errors at intersections
· 23% of drivers made signaling errors during lane changes
· 57% of drivers did not fully stop at stop sign controlled intersections
· 13% of drivers did not follow prescribed route
· 30% of drivers did not adequately scan
· 37% of drivers sped
· 17% of drivers made critical errors
3.4.6 Results: Open Road to Walnut Creek
Three individuals made critical errors during the Walnut Creek portion of the test. Two of the
three individuals only made critical errors in the Walnut Creek section, while one participant
made critical errors in both Rossmoor and Walnut Creek. Two of the three errors were failing
to stop the vehicle at a stop sign before making a right hand turn. Both drivers failed to slow
the vehicle, come to a complete stop behind the crosswalk, yield the right of way, or scan
appropriately at the intersection. Both drivers never slowed down to below approximately 20
mph before making the turn.
The third critical error was by far the most dangerous of the entire study. This driver ran a red
light in downtown Walnut Creek and was completely unaware that he had done so. The light
had turned red well before the driver approached the intersection. Examination of the video
focused on the driver’s face showed absolutely no hesitation or awareness that the driver had
just driven through the red light.
The majority of non- critical errors made by most drivers mimicked those observed during the
57
Rossmoor section of the course. Head turning errors ( not turning head appropriately to scan
and/ or not checking blind spot) were the most frequent, particularly at intersections and during
lane changes. A little less than half of the drivers demonstrated generally inadequate scanning
behavior during the Walnut Creek section of the course. Another type of error made
frequently was turning too wide at an intersection. Once the driver had turned, they often did
not stay in the appropriate lane ( they often turned and “ drifted over” into the other lane).
Key points from Walnut Creek section include:
· 73% of drivers made head turning errors at intersections
· 77% of drivers made head turning errors during lane changes
· 20% drivers made head turning errors while parking
· 63% of drivers turned too wide
· 17% of drivers failed to signal at intersections
· 17% of drivers failed to signal before changing lanes
· 23% of drivers failed to signal during parking/ pulling out
· 20% of drivers rolled through stop signs
· 43% of drivers inadequately scanned during drive
· 17% of drivers sped during drive
· 17% failed to have two hands on wheel during all of drive
· One driver performed a self- distracting activity while driving ( looking at map, misses
light turning green)
· 10% of drivers committed critical errors
3.4.7 Results: Overall Driving
Overall, seven participants would have failed a DMV road test because they made critical
errors in Rossmoor and/ or Walnut Creek. Additionally, our driving performance evaluators
scored those seven participants as well as one additional participant as “ people they would not
ride with in a vehicle” due to unsafe driving behaviors.
58
Two interesting observations were made in a number of participants with respect to usability
issues. First, 20% of participants rested their hands during driving on the central steering
wheel spokes instead of gripping the wheel itself. This seemed like an odd hand placement,
and potentially unsafe if the airbag were to deploy. Additionally, hand placement on the
spokes would increase the force required to actually turn the wheel. The second observation
of note was that several participants ( 20%) frequently utilized tissues during the course of
driving. This was sometimes a distracting activity because they would have to reach for the
tissues and did not seem to have an adequate place to store and dispose of the tissues.
The most frequent errors that were made by nearly all drivers were related to head turning and
scanning activities. This was not surprising, given the number of participants with limited
neck and torso flexibility and decreased visual search and divided attention abilities. A
logistic regression model was used to determine which intake examination measures were
associated with head turning errors. The main predictor of head turning errors at intersections
was failing the seated head turning task during the intake examination. Those drivers who
could not identify an object within five seconds on a computer screen placed ten feet away
directly behind them, had a 5.6- fold increased risk of making head turning errors at
intersections. Those drivers who failed to look fully through their rear window before backing
out of the parking space, had significantly less neck range of motion compared to those who
did look appropriately ( Mean neck rotation available for those who turned appropriately = 64 ° ;
Mean neck rotation available for those who did not turn appropriately = 56 ° ; p = 0.046).
3.4.8 Results: Ingress/ Egress and Loading of Packages
Individual ingress/ egress performance and loading of packages for all drivers is compiled in
Appendix B. The majority of participants loaded both the suitcase and the grocery bag into the
trunk. The drivers who did not use the trunk typically placed the items on the floor of the
backseat. No one used the front passenger side to load packages.
59
Participants had the least difficulty getting into the driver seat. Getting out of the driver seat
and into the rear passenger seat were the next most difficult ingress/ egress tasks. Nearly all
participants had some difficulty or used altered strategies compared to young adults when
getting out of the rear passenger seat ( 91%).
Key results include:
Suitcase Loading
· 70% placed the suitcase in the trunk
· 21% placed the suitcase on backseat floor
· 9% placed the suitcase on backseat
Grocery Bag Loading
· 64% placed the groceries in the trunk
· 21% placed the groceries on the backseat floor
· 15% placed the groceries on backseat
Ingress
· 28% had difficulties getting into the driver seat
· 67% had difficulties getting out of the driver seat
· 65% had difficulties getting into rear passenger seat
· 91% had difficulties getting out of rear passenger seat
· Required the use of one arm/ hand during ingress - driver seat = 12%, backseat = 32%
· Required the use of one arm/ hand during egress - driver seat = 24%, backseat = 23%
· Required the use of two arms/ hands during ingress - driver seat = one person, backseat
= 9%
· Required the use of two arms/ hands during egress - driver seat = 9%, back seat = 14%
3.4.9 Limitations of the Study
A major limitation was the use of a relatively small and high functioning convenience sample,
60
which limits the power and external validity of the study. Unfortunately, given the risks
involved with conducting an open- road driving study and the large amount of time needed for
data analysis, our options were limited. Although the video technology allowed us to perform
a less intrusive assessment of driving performance, knowledge of the equipment may have
affected performance. Use of the subject’s own vehicle allows the driver to perform in a
naturalistic setting, but does not allow for a standardized view from the video cameras.
Similarly, use a non- standardized driving route in Walnut Creek and at different times of day
meant that subjects may have encountered different driving situations.
4.0 CONDUCT DRIVING EXPERIMENTS
4.1 Introduction
In Section 2.0, the case is presented that older drivers are over represented in LTAP/ OD
crashes ( left turn across path with opposite direction traffic). Specifically, older drivers may
have difficulty judging the speed of other vehicles and available time to turn in front of
oncoming vehicles.
One possible solution conceptualized at California PATH is an in- vehicle message for a
LTAP/ OD gap advisor. This stems from research conducted under the Intersection Decision
Support ( IDS) project, conducted under the auspices of the Infrastructure Consortium ( IC).
The IC is comprised of the US Department of Transportation ( DOT), California DOT
( Caltrans), Minnesota DOT, and Virginia DOT. The IDS project addresses the application of
infrastructure- based and infrastructure- vehicle cooperative systems to address intersection
safety and is the predecessor to the US DOT, Infrastructure Consortium and Collision
Avoidance Metrics Partnership ( CAMP) Cooperative Intersection Collision Avoidance System
( CICAS). ( For more information on CICAS, see the second initiative under
< http:// www. its. dot. gov/ press/ Initiatives4. htm>.) PATH is a research participant in both the
IDS and fledgling CICAS programs and the institution most focused on LTAP/ OD.
61
In IDS, our emphasis has been LTAP/ OD warning from the infrastructure. However, in
concepting alternate messages to make left turns even more safe for older drivers, we have
considered a more salient on- board message. This gives rise to the LTAP/ OD display used for
Toyota GapAdvise.
How would such a system work? The subject vehicle ( SV) – or the vehicle equipped with the
Toyota GapAdvise LTAP/ OD warning system – approaches the intersection. It has a
( permissive) green signal, but there is no left turn arrow or protected cycle, so the driver slows
down to a stop to check if it is safe to make a left turn onto at the intersection. The SV driver
may be older or otherwise not able to easily judge the speed or location of this approaching
traffic, making it hard to decide whether or not to turn. While the SV driver is trying to
determine whether the left turn is safe, other vehicles (“ Principal Other Vehicles” – POV) are
approaching the intersection with the intent of proceeding straight. Therefore, intermittent
gaps, some safe and some not safe may be present.
In order to help the SV driver prevent a collision or near collision, the PATH IDS system
issues a warning to the SV driver by illuminating the dynamic “ no left turn” sign – or the
Toyota GapAdvise LTAP warning system provides a similar in- vehicle warning. These are the
alternatives we studied in this task.
4.2 Research Questions
In exploring the concept of an in- vehicle gap advice system, this study addressed the following
four research questions:
1. What is considered an unsafe gap?
2. When should you give the warning to be effective in influencing the drivers’ decisions?
3. How should the warning be given?
4. How effective might the system be in reducing the number of unsafe turns?
In order to define what an unsafe gap is, we must first discuss how to measure gap. The term
gap ( either measured in distance or time) is most often used in the literature to refer to the
space between the rear bumper of one vehicle and the front bumper of the next where the
62
vehicles are traveling in the same direction. Thus, from the turning vehicle’s point of view,
there could only be a gap in traffic between two oncoming vehicles. While this is the case
sometimes, it cannot be used to describe all possible cases experienced while driving.
Occasionally the term lag ( again either in terms of time or distance) has been used in the
literature to describe the space between the front bumper of the turning vehicle and the front
bumper of an approaching vehicle. Finally, from an intersection- centric point of view, all
vehicle movements might be described in terms of t2i ( time to intersection) or d2i ( distance to
intersection).
Unfortunately, none of these terms adequately describe the nuances associated with having
two moving vehicles. For example, if we were to describe the vehicle movements in terms of
lag, the value and interpretation changes as the vehicles approach the intersection. Thus, a lag
of 3 seconds where the turning vehicle is already at the intersection is entirely different than a
lag of 3 seconds where both vehicles are still 1.5 seconds away from the intersection. To
eliminate this problem, we introduced the concept of trailing buffer. The trailing buffer
roughly equates to a measure of spare time. Assuming the turning vehicle is going to complete
its turn in front of the oncoming vehicle, how much spare time would remain before the
oncoming vehicle reached the intersection? Given the very preliminary and conceptual nature
of this study, trailing buffer was intended to be studied from the range of nobody would turn in
front of the approaching traffic to everybody would turn.
The second research topic relates to the question of decision point. At some point during the
approach of the turning vehicle, the driver must decide whether there is time to turn, or
whether s/ he must stop at the intersection and wait for the approaching traffic to clear. Any
advice or alert given by a system should coincide with this decision making process.
Warnings that come too late carry the risk of being ignored because the driver has already
committed to the turn and might not have time to integrate the warning and change his or her
behavior. Warnings that come too soon might be seen as a nuisance, especially if the driver
disagrees with the system’s assessment of the situation.
63
Ongoing PATH research14 has examined the decision point issue by observing drivers making
left turns in an urban environment setting. As shown in Figure 4- 1, as the turning vehicle
enters the left turn lane, it is impossible to tell whether that vehicle will turn without stopping,
or stop and then turn based on the speed trajectory alone. However, around 20- 25 meters from
the stop bar, two clusters of speed trajectories become noticeable: those that intend to stop
( Trajectory 4), and those that intend to turn without stopping ( Trajectory 1). This evidence
suggests that the decision point lies in the range of 20- 30 m from the stop bar.
Figure 4- 1. Intersection Approaches: Turned Without Stopping vs. Stopped Before Turn.
The final research topics, how to implement the in- vehicle warning and how effective such a
warning might be, were not intended to be the primary focus of this study, but they are
nonetheless addressed by virtue of creating and testing a prototype gap advice system.
14 Cody, D. ( 2004). Intermediate summary of IDS ( intersection decision support) field test results. Presented at
the IDS Quarterly Meeting 9/ 26- 9/ 29 in Minneapolis, MN. Berkeley, CA: California PATH.
Trajectory 1 Trajectory 4
Left turn lane Stop Bar Middle of Intersection
64
4.3.1 Test Plan
4.3.2 Overview
The goal of this experiment was to observe driver LTAP/ OD behavior with the introduction of
a conceptual in- vehicle gap advice warning system. The conceptual system would evaluate
the speeds and distances of the vehicles approaching the intersection and provide an alert to
the driver if it was deemed unsafe to make a left turn in front of the oncoming vehicle. During
the experiment, the SV approached the intersection at approximately 20 mph with instructions
to make an unprotected left turn at the intersection ( i. e., the SV has a green light but must yield
the right of way to oncoming traffic). The POV approached the intersection from the opposite
direction at approximately 25 mph. The arrival of the vehicles ( the available gap to turn in
front of the POV) and the timing of the warnings were varied in the experiment.
4.3.2 Test Participants
Twenty licensed drivers in two age groups, ten younger ( 20 to 38 years old, mean of 28.3) and
ten older ( 65 to 84 years old, mean of 75.2), participated in this experiment. Within each age
group, there were five men and five women drivers. Participants were recruited through email
advertisements placed on various UC Berkeley student mailing lists and a “ Resource Center on
Aging” ( see < http:// ist- socrates. berkeley. edu/~ aging/>) monthly newsletter. There was no
overlap between the test participants in the focus group and the participants in this test. All
subjects were paid a nominal $ 30 for their participation regardless of their performance in the
experiment.
Based on the responses to a background questionnaire, the majority of the test participants
regularly drove small to midsized sedans or wagons, such as the Toyota Corolla or Honda
Accord. Ten percent of the participants drove small SUV’s, such as the Honda Element or
Suburu Forester, and twenty percent drove larger cars such as the Buick Century or VW
Passat. As shown in Table 1, younger drivers reported driving less than 5000 miles per year
65
more often than older driver, which most likely reflects the younger driver sample population
being weighted towards urban university graduate students.
Table 4- 1. Annual mileage
Annual Mileage Younger Older
< 5000 40% 20%
5000 - 10,000 40% 40%
> 10,000 20% 40%
As shown in Table 4- 2, most of the driving time for younger drivers was spent on freeways,
with the rest of the time split between urban and suburban settings. Older drivers were more
varied, spending most of their time in urban driving. Neither age group spent much time on
rural roads. Overall, these results are not inconsistent with the mix of roads in the San
Francisco Bay Area.
Table 4- 2. Driving habits by driving environment
Younger Older
Female Male Mean Female Male Mean
Freeways 52% 42% 47% 41% 21% 31%
Urban 20% 34% 27% 45% 45% 45%
Suburban 20% 18% 19% 9% 31% 21%
Rural 8% 8% 8% 4% 3% 3%
Tables 4- 3 and 4- 4 show the mix of day vs. night driving and familiar vs. unfamiliar
destinations. Younger drivers reported slightly more night driving with a mean of 40 percent
of their time spent behind the wheel at night, while the older drivers only averaged 30 percent.
Similarly, younger drivers were also more apt to visit unfamiliar destinations, than were older
drivers.
Table 4- 3. Driving habits by time of day
Younger Older
Female Male Mean Female Male Mean
Day 57% 63% 60% 61% 78% 69%
Night 43% 37% 40% 39% 22% 31%
66
Table 4- 4. Driving habits by destination
Younger Older
Female Male Mean Female Male Mean
Familiar 79% 59% 69% 81% 77% 79%
Unfamiliar 21% 41% 31% 19% 23% 21%
About 40 percent of older drivers and 70 percent of younger drivers reported that urban/ city
driving was “ sometimes difficult.” Regarding the factors that cause the most difficulty in
driving, 65 percent reported “ other drivers,” 45 percent reported “ intersection complexity,”
and 35 percent reported “ pedestrians.” Left turn across path with opposite direction traffic and
freeway merging were most often reported as the most difficult driving maneuvers when it
came to estimating vehicle speed.
4.3.3 Experiment Design
Overview
Two factors were manipulated in this experiment. The first factor manipulated was the arrival
of the SV and POV to the intersection, which translates time available for the SV to turn in
front of the POV. The second factor manipulated was the warning timing, the point during the
SV’s approach to the intersection, at which, the warning was given. The speed of the SV was
fixed at 20 mph and the speed of the POV was fixed at 25 mph; however, as both of these
speeds were human controlled, variations were expected between trials. The mean SV
approach speed was 20.5 mph ranging from 16 to 29 mph. The mean POV approach speed
was 24.4 mph ranging from 21 to 29 mph.
Trailing Buffer ( Spare Time)
The arrival to the intersection of both the SV and POV were described using the concept of
trailing buffer measured in seconds. This calculation roughly equates to a theoretical
projection of how much spare time would remain if the SV made a typical turn in front of the
POV. Thus for any given SV position, the predicted trailing buffer could be calculated by
67
subtracting the SV time to clear the intersection from the POV t2i. In this calculation it is
assumed that the POV will maintain its current speed. Likewise, the SV will maintain its
current speed until it decelerates to a turning speed, then continue through the intersection at
its turning speed. A regression of trials at the RFS intersection showed that the typical SV
turning speed was 13.18 mph ( 5.89 m/ s), and the typical deceleration rate was 0.16 g
( 1.61 m/ s/ s). Using this model, the typical turning time for the RFS intersection ( the time from
SV d2i equals zero to the time the SV rear bumper clears the intersection) was predicted at
2.85 s.
In interpreting the trailing buffer, a positive value
( Figure 4- 2) would indicate that the SV’s rear bumper
cleared the intersection before the arrival of the POV.
For a nominal POV speed of 25 mph and a 10- meter
wide intersection, a trailing buffer between - 3.5 and
0 seconds would indicate a very close call or a
potential collision. Trailing buffers less than about
- 3.5 seconds would indicate that the POV cleared the
intersection before the SV’s arrival. For this
experiment, three nominal target trailing buffers, - 1.5,
- 0.5, and 0.5 seconds, were used
Figure 4- 2. Positive trailing buffer.
Warning Timing
There were four conditions relating to the warning timing used in the experiment. First, there
was the possibility that no warning would be given on a particular trial. Otherwise, warnings
were given in terms of three SV distances to intersection stop bar ( outer crosswalk line): 16,
24, or 32 m. At an SV speed of 20 mph, these values roughly translated to 2, 3, and 4 seconds
to the intersection stop bar.
68
Summary
A total of four practice trials and twenty- four test conditions or intersection approaches were
completed for each driver. Table 4- 5 shows the number of trials for each combination of
warning point and target trailing buffer. A warning was not shown when the trailing buffer
value was equal to or exceeded 0.5 seconds as this was almost universally considered a safe
turning condition in pilot testing. Similarly, a warning was always shown when the predicted
trailing buffer was less than - 1.5 seconds.
Table 4- 5. Number of trials for each test condition.
Trailing Buffer Warning Point
16 m 24 m 32 m No Warning
- 1.5 s 3 3 3 0
- 0.5 s 3 3 3 3
0.5 s 0 0 0 3
4.3.4 Test Materials and Equipment
Test Vehicles
The test participants drove the California PATH instrumented Ford Taurus sedan, model year
1998 ( see Figure 4- 3), which was designated at the SV, or the vehicle making the left turn at
the intersection. The POV was a white 1996 Buick LeSabre, driven by a confederate driver.
The Taurus was outfitted with a video recording system, a vehicle data recording system, a
laptop dedicated to the DVI ( driver- vehicle interface), and an off- head, video- based FaceLab
eye tracking system ( running software version 3). However, the only instrumentation visible
to the driver were the two cameras mounted on the dashboard for the eye tracking system, and
the display used for the DVI.
69
Figure 4- 3. California PATH instrumented Ford Taurus sedan.
The DVI used to display the in- vehicle warnings was a 7” LCD display ( Xenarc Model
700YV), mounted in the high center position as shown in Figure 4- 4 in an attempt to
approximate the position of a typical navigation system display. The no- left- turn sign shown
on the screen for the visual warning had the characteristic of looming, i. e., the red circle and
slash portion of the graphic increased and decreased in width by about 20 percent at a rate of
about 2 Hz. This gave the impression of a flashing effect, helping to attract attention to the
display, without ever having the no- left- turn warning disappear. The audio portion of the DVI
was played through the displays speaker with the volume adjusted to a comfortable level for
each driver. The sound used to indicate an unsafe gap was a pair of 2000 Hz tones at a 200 ms
cadence. All of the information displayed on the Taurus DVI was received via an 802.11b
wireless link from the infrastructure. The vehicle- based sensors, such as the radars, were not
used to calculate or display warnings on the DVI.
70
Figure 4- 4. DVI mounted in the Taurus displaying the No- Left- Turn Warning.
Test Intersection
The experiment was run at the UC Berkeley, RFS Intelligent Intersection. This intersection is
a typical four- leg intersection with one lane in each direction ( no left or right turn lanes). The
approach from the POV direction was approximately 1000 meters, while the approach from
the SV direction was approximately 100 meters. Using a suite of in- pavement magnetic loops,
3M microloops, and EVT- 300 radars, and 802.11b wireless links to the vehicles, a roadside
PC- 104 monitored the SV and POV speed, distance, and acceleration continuously during each
trial. The roadside PC- 104 then rebroadcast the information along with a determination of any
warning conditions to the SV over the 802.11b wireless link. The traffic signal was kept in the
green phase for the SV and POV throughout each trial.
71
4.3.5 Experimental Protocol
Test Activities and Sequencing
Upon the arrival of the test participant, s/ he was greeted and asked to read and sign a consent
form and fill out a background questionnaire ( both in Appendix C). They were then seated in
the instrumented Taurus and asked to adjust the seat, mirrors, and steering wheel to a
comfortable position. The eye tracking system was calibrated for the driver, and the sequence
of the experiment was explained step- by- step in detail to the driver ( see Table 4- 6).
Throughout the experiment, the experimenter sat in the rear passenger seat of the Taurus.
The arrival of the vehicles at the intersection ( and subsequent trailing buffer) was manipulated
by adjusting the start time of the SV relative to the start time of the POV, which was controlled
by the roadside PC/ 104 computer stack. The POV driver started each trial by sending a signal
to the roadside computer, which in turn, started a countdown, sending a start signal to each
driver at the appropriate time. To the SV driver, the start signal seemed to come at a random
time between 10 and 15 seconds after the experimenter radioed that the SV was in position and
ready. The trial was considered completed after the test participant completed the left turn.
72
Table 4- 6. Typical trial sequence.
Activity Sequence Driver Instruction DVI
1. Line up vehicles The test participant parks the SV
approximately 80 m from the
intersection and waits for the start
signal. ( The POV parks 260 m
from the intersection.)
2. Safety check The experimenter radios that SV is
in position and ready to start when
the track is clear.
3. POV driver starts
the trial
The POV driver initiates the start
of the trial by sending a signal over
the wireless network to the
roadside PC- 104.
4. POV receives the
start signal
The POV driver accelerates up to
25 mph towards the intersection.
5. SV receives the
start signal
Upon hearing the phrase “ Left
Turn Ahead” spoken by the DVI,
the test participant was instructed
to accelerate to 20 mph, drive up to
the intersection, and make a left
turn.
Audio: “ Left Turn Ahead.”
5. SV receives the
unsafe gap alert
At the designated warning point
for the trial, the SV displayed a
warning based on the trailing
buffer. The DVI unsafe gap
warning screen change was
preceded by “ beep beep” sound to
alert the driver. The warning
screen consisted of a looming no-left-
turn sign and a countdown bar
representing the POV distance to
intersection.
Audio: “ Beep Beep”
6. Trial completed After the SV has made its left turn,
the trial was completed, and the
experiment asked probing
questions about the trial.
( Same as Activities 1- 4)
73
Practice Trials and Instructions to Drivers
The test participants were instructed to approach the intersection at 20 mph and make a left
turn as they would normally. They were instructed to turn in front of the oncoming vehicle if
they felt it was safe and appropriate, whether or not a warning was present. Warnings were to
be treated as advice. The test participants were also discouraged from speeding up faster than
20 mph in order to beat the oncoming vehicle.
Four practice trials were given before the start of the test. The first two practice trials were
given without the DVI unsafe gap alert, simply to familiarize the drivers with the trial
protocol, the intersection layout, and the handling of the Taurus. The second two practice
trials added the concept of DVI warnings. Both the warning and its meaning were described to
the driver a priori, and thus, the drivers were not required to blindly interpret the meaning of
the device.
Post- Trial Probing Questions
After each trial, the test participant was asked two probing questions by the experimenter.
1. Did you think there was enough time to turn in front of that car?
Responses were coded as follows:
a. Driver answered yes, and turned in front of the POV.
b. Driver answered yes, but stopped to let the POV pass.
c. Driver answered maybe, if s/ he was in a hurry, but stopped to let the POV pass.
d. Driver answered no, and stopped to let the POV pass.
2. When the warning came, did you feel it was early, late…?
Responses were coded on a scale of 1- 5 with 1 being too early, 5 too late, and 3 just right.
74
4.4 Results
4.4.1 Trailing Buffer
For each trial or intersection approach, there were two possible outcomes, the driver could turn
in front of the oncoming vehicle or stop and wait for it to pass. If the driver chose to stop, an
opinion was solicited as to whether the driver thought there was enough time to turn after the
fact. Figure 4- 5 depicts these results broken down by half- second increments of trailing
buffer. Thus, when the trailing buffer was greater than 1.0 seconds, almost all drivers turned
in front of the oncoming car. When the trailing buffer was between - 1.0 and - 0.5 seconds, 40
percent of the time, drivers thought there was not enough time to turn; and 60 percent of the
time, drivers thought there was enough time to turn. However, the turn was actually only
made a little less than 30 percent of the time.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
- 3.5*
- 3.0
- 2.5
- 2.0
- 1.5
- 1.0
- 0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0*
3.5*
4.0+*
Average Predicted Trailing Buffer ( sec)
Stopped: Not Enough Time to Turn
Stopped: Might Turn if in a Hurry
Stopped: But Could Have Turned
SV Turned
Figure 4- 5. Decision to turn as a function of trailing buffer.
As shown in contrasting Figures 4- 6 and 4- 7, younger drivers were slightly more aggressive
than older drivers with a higher percentage of turns being made in the - 2.0 to the - 0.5 second
range. However in the - 0.5 to 0.0 second trailing buffer range, older drivers made the turn
75
more than 50 percent of the time, while younger drivers made the turn only about 35 percent
of the time.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
- 3.5
- 3.0
- 2.5
- 2.0
- 1.5
- 1.0
- 0.5
0.0
0.5
1.0
1.5
2.0*
2.5*
3.0*
3.5*
4.0+*
Average Predicted Trailing Buffer ( sec)
Stopped: Not Enough Time to Turn
Stopped: Might Turn if in a Hurry
Stopped: But Could Have Turned
SV Turned
Figure 4- 6. Decision to turn as a function of trailing buffer for younger drivers.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
- 3.5*
- 3.0
- 2.5
- 2.0
- 1.5
- 1.0
- 0.5
0.0
0.5
1.0
1.5
2.0*
2.5*
3.0
3.5*
4.0+
Average Predicted Trailing Buffer ( sec)
Stopped: Not Enough Time to Turn
Stopped: Might Turn if in a Hurry
Stopped: But Could Have Turned
SV Turned
Figure 4- 7. De