Proving Test Report: Advanced Cruise-Assist Highway System Proving Tests
|
 |
Hiroyuki Mizutani Manager Advanced Cruise-Assist Highway System Research Association |
1. Introduction
The Advanced Cruise-Assist Highway Systems (AHS) function to improve the safety
of road traffic by using road infrastructure sensors to detect various dangerous
conditions that drivers and vehicle-side sensors cannot detect on their own,
and using communications technology to provide that information to the drivers.
Research and development for practical application of AHS has been proceeded
according to the Cruise Assist Systems Infrastructure Requirements (Primary),
formulated in October 1999. These proving tests of the element technologies
and systems that had been the subject of research and development have finally
brought this program to the point of embarking on the process of practical application.
The proving tests were carried out as joint tests of AHS and the Advanced Safety
Vehicle (ASV), which combine to form the Cruise Assist Systems for cooperative
vehicle-highway. The testing was also intended to build a consensus both inside
and outside Japan, to promote understanding, and to conduct multi-faceted evaluations
of AHS infrastructure. Consequently, outside applicants for participation in
joint research were sought, and they did take part in the testing. This report
will take up that portion of last year's tests assigned to AHS, and will introduce
testing plans for the field operation tests that are scheduled to begin this
year.
2. Overview of Proving Tests
2.1 Test Objectives and Time Frame
The test objectives were (1) to verify the functions, performance, and feasibility
of Cruise Assist Systems, which are based on cooperative vehicle-highway through
the linking of ASV and AHS, (2) to build a consensus on this program both inside
and outside Japan, and (3) to promote understanding of the program.
The tests were conducted mainly from October to December 2000. Public demonstrations
known as Joint Tests-Demo 2000 were also conducted during that time. These
involved opening proving tests to the public, allowing people to take test rides,
and holding events such as workshops, panel sessions, and exhibitions (Fig.
1.).
| Fig. 1. Overview of Proving Tests |
 |
2.2 Test Facilities
The tests were conducted on the test course of the National Institute for Land
and Infrastructure Management (formerly known as the Public Works Research Institute)
(Fig. 2.).
| Fig. 2. Test Course and Test Applications |
 |
2.3 Test Participants
The test participants included the AHSRA, the ASV developers (thirteen domestic
automobile manufacturers), and seven Japanese and foreign organizations that
took part in outside applicant testing (Fig. 3.). The cooperative ASV-AHS proving
tests were conducted jointly by the former Ministry of Construction and the
former Ministry of Transport. The very fact that tests took place involving
the coordination of vehicles and infrastructure, and aiming toward practical
application, was of great value in itself.
| Fig. 3. Overview of Proving Tests |
 |
3. Proving Test Results
Key points in the results obtained from the AHS proving tests are described
below. The tests covered the validity of infrastructure system design figures
(verification of requirements and systems), the effectiveness of services, driver
acceptance, and the performance of element technologies (Fig.4.)
| Fig. 4. Test Evaluation Items |
 |
3.1 Validity of Infrastructure System Design figures (Verification of Requirements and Systems)
| (1) |
Driver Response Time to Information Provision One of the hypothetical values
set in the requirements is the driver's response time to information provision.
This figure affects installation intervals and other design factors for sensors
and beacons on the infrastructure side. The requirements took a hypothetical
value of 2.65 seconds for driver response time, and the validity of this value
was tested in actual driving by test subjects. The testing was conducted after
familiarizing the test subjects with the nature of the services involved by
means of reference materials, videos, and so on. The test results showed that 3 seconds was better than the figure of 2.65 seconds set in the requirements. Some test subjects, in fact, had response times even longer than 3 seconds, but they were able to come to full stops before the obstacles that were placed on the test course. This shows that the subjects adjusted their rate of deceleration after their response to information in order to stop. As a result, the actual design must take into account the combination of response time with deceleration (Fig.5.). |
| Fig. 5. Driver Response Time to Information Provision |
 |
| It is important to note these tests were conducted within the limited space of the test course. Consequently, the subjects easily grew familiar with the road shapes on the curves used to test services related to over shooting, and would know in advance that they were approaching a curve so they could adjust their driving accordingly. It will be necessary, therefore, to conduct further testing at other locations with natural curves by means of field operation tests. | |
| (2) | Driver Deceleration in Response to Information Provision The requirements also set a hypothetical value for deceleration. The tests acquired data on this together with response time data. Verification of the requirements figure showed that deceleration following reception of services should be 1.5-2.5 m/s2. The requirements had set a value of 3 m/s2, but this was determined to be slightly too sharp. The proper values for response time and deceleration will be determined in the future through consultation with automobile manufacturers. (Fig. 6.) |
| Fig. 6. Deceleration in Response to Information Provision |
 |
| (3) |
Evaluation of Lane Marker Placement Validity The validity of lane marker placement was evaluated by carrying out tests of lane departure services with varying longitudinal placement of markers. The service to support prevention of lane departure will not be described in detail here. The tests were arranged so that an alarm would sound when a driver was left a lane, upon which the driver would attempt to return to the lane. The tests measured that response time and the yaw rate to return. The working hypothesis was that it would be necessary to sound the alarm when displacement from the lane center was 1.1 m on curves and 0.8 m on straight sections, in order to avoid danger. The testing verified whether the alarm was issued properly and at the appropriate location when the placement of the lane markers was changed. It had already been determined by simulations that instability resulted when lane markers were placed at intervals greater than 8 m. Therefore, these tests verified the actual conditions with lane marker intervals within the zone of stability, at 8 m and less. It had been assumed that lane markers would be installed at intervals of 2 m, and testing with that arrangement was conducted by having the sensors skip reading of some of the markers. However, the results showed that under given consistent conditions, the arrangement of markers in longitudinal series was sufficient to provide services when the longitudinal interval between markers was under 4 m on curves and under 8 m on straight sections. As course conditions and test time limitations imposed certain restrictions, it will be necessary to run further, more complete tests in order to obtain definitive values. |
3.2 Effectiveness of Services
| (1) |
Comparison of driver behavior by differences between presence and absence
of support for prevention of collisions with forward obstacles Here the example of services on an ordinary road to support prevention of collisions with forward obstacles will be used (Fig. 7.). |
| Fig. 7. Analysis of Maximum Deceleration with Prevention of Collisions with Forward Obstacles |
 |
| In this graph, the horizontal axis shows the maximum deceleration applied by test subjects while driving. The vertical axis shows the number of instances. The solid line shows the distribution of instances when services were provided, and the broken line the distribution of instances where services were not provided. The graph shows that the maximum deceleration applied by drivers when services were provided was lower than when services were not provided by an average of approximately 0.6 m/s2. This indicates that the reception of information by drivers reduced the number of cases in which drivers hastily applied emergency deceleration after detecting an obstacle themselves. It is particularly noteworthy that the number of test subjects whose deceleration exceeded 5.0 m/s2, which is defined as emergency deceleration, decreased by 95%. In other words, on the presumption that occasions where deceleration is greater than 5.0 m/s2 are more likely to result in accidents, the probability is reduced by 95%. These data can, therefore, be understood to indicate that support for prevention of collisions with forward obstacles is effective in preventing such accidents. There were virtually no differences related to the age of test subjects. The provision of services clearly reduces the number of test subjects who exceed the maximum deceleration of 5 m/s2. Those are the subjects in the region enclosed by the dark line (Fig. 8.). A deceleration of 5 m/s2 is equivalent to heavy braking, and the reduction of heavy braking itself signifies the probability of correspondingly safer driving. |
| Fig. 8. Distribution of Maximum Deceleration with Support for Prevention of Collisions with Forward Obstacles (by Age) |
 |
| (2) |
Comparison of driver behavior by differences between presence and absence
of support for prevention of over shooting on curves An example of support for prevention of over shooting on curves will be used here (Fig. 9.). The horizontal axis of this graph shows the speed at which test subjects entered the curve. The vertical axis shows the number of instances. The solid line shows the distribution of instances when services were provided, and the broken line the distribution of instances where services were not provided. The graph indicates that the drivers' speed of entry into the curve when service for prevention of over shooting was provided was lower than when that service was not provided. This indicates that reception by drivers of the information provided actually caused them to reduce their driving speed in advance, and this allowed a larger proportion of them an adequate margin of time in which to enter the curve. Verification results obtained from use of a driving simulator with the same conditions as the present tests (curve radius of 150 m) showed that speeds of 100 km/h and above result in a higher probability of over shooting, spins, or other dangers. Here, the number of test subjects who entered curves at speeds of 100 km/h or higher declined by 70%. Here, the number of test subjects who entered curves at speeds of 100 km/h or higher declined by 70%, thus reducing that probability by 70%. These data can, therefore, be understood to indicate that support for prevention of over shooting on curves is effective in preventing such accidents. |
| Fig. 9. Distribution of Curve Entry Speeds with Support for Prevention of Over Shooting on Curves |
 |
| (3) |
Comparison of driver behavior by differences between presence and absence
of support for road surface condition information for maintaining headway, etc.
An example of support for road surface condition information for maintaining headway, etc. will be given here. (Fig. 10.) The test is conducted by having two cars drive through an area where water film has formed. The lead vehicle deliberately decelerates before reaching the area of water film. The driver of the following vehicle will tend to decelerate when the headway reaches a point the driver thinks is dangerous. There is a difference, however, between the timing of braking when the driver knows in advance that water film lies ahead, and the driver's timing for braking when that information has not been provided. This test measures that difference in timing. The solid line shows the distribution of instances when services were provided, and the broken line the distribution of instances where services were not provided. It is evident that, with this service, the headway at the point when the following vehicle applies the brakes increases by an average of approximately 0.65 seconds, from 1.8 to 2.45 seconds. The provision of road surface information, whether regarding dry or water film conditions, allows the following vehicle to apply the brakes sooner than when that information is not provided. This tends to increase headway, and so these data can be understood to indicate that support for road surface condition information for maintaining headway, etc. is effective in preventing rear-end collisions with the lead car and other such dangers. |
| Fig. 10. Distribution of Headway with Support for Road Surface Condition Information for Maintaining Headway, etc. |
 |
| (4) | Effectiveness of Other Services Other services than those described above were also tested, such as support for prevention of crossing collisions, support for prevention of right turn collisions, support for prevention of collisions with pedestrians crossing streets, and support for prevention of lane departure. The effectiveness of these services was verified by means of questionnaire surveys of the test subjects. Though they showed some variation, from 73% to 92% of test subjects responded that they found these services useful. With respect to support for prevention of lane departure, in particular, 100% of female test subjects and subjects aged 65 and over responded that it was useful. (Fig. 11.) |
| Fig. 11. Summary of Findings on the Effectiveness of Services |
 |
3.3 Driver Acceptance
| (1) |
Evaluation of the sense of security resulting when services are provided
The questionnaire survey sought to determine the driver's sense of security when services are provided compared to when they are not. The results showed that, apart from support for prevention of collisions with forward obstacles provided in rain or fog conditions, over 70% of test subjects responded that provision of services gave them "a greater sense of security." This confirmed that the services are accepted by most drivers. The services were somewhat less accepted in rain and fog because the lack of visibility made for a very harsh driving environment, and the services that were provided under those test conditions were not in themselves sufficient to make the subjects feel fully secure (Fig. 12.). |
| Fig. 12. Evaluation of Driver Sense of Security when Services are Provided |
 |
| (2) |
Recognition and utilization of information content provided by services The questionnaire survey was also used to determine whether drivers recognized the content of information provided by services. The results showed that the recognition rate was high except in rain or fog conditions. The rate is thought to have been lower in a rain or fog environment because the poor visibility kept the drivers concentrated on their driving, causing them to miss hearing the voice guidance information or to fail to understand it. With regard to utilization of the information provided, drivers acknowledged receiving that information at intersections and other locations. However, some drivers claimed that they decelerated before intersections by their own judgement rather than because of the information they received. More drivers expressed this view with regard to information provision at intersections than for other locations.. (Fig. 13.) |
| Fig. 13. Summary of Driver Acceptance |
 |
3.4 Verification of Element Technology Performance
Separate verification of element technologies was implemented on actual roads.
This report, however, deals with test course issues, and here it will present
various problems that arose through use of the ASV. Issues that arose in connection
with road-to-vehicle communications include interference with signals, the effects
of buildings and other structures in the vicinity, and the temperature of on-board
radio equipment. In connection with the content of information provided, the
fact that road-to-vehicle communications have a certain limited capacity meant
that information sometimes ended up exceeding that capacity. The accuracy of
position information from lane markers also sometimes fluctuated suddenly. Road
sensors experienced detection failures and detection errors. Specifically, the
sensors sometimes failed to detect targets when there were moving cloud shadows,
when trees swayed, when it was raining, and when it was cold.
Work is currently underway to isolate issues of this kind in order to resolve
them. We intend to study specific measures for dealing with these issues, and
to incorporate them into our future research.
3.5 Results of Questionnaire Survey during Public Demonstration
A questionnaire survey was conducted of participants in Joint Tests Demo2000,
a four-day series of public demonstration held at Tsukuba City in Ibaraki Prefecture
from November 28 to December 1, 2000. The results indicate strong interest in
Cruise Assist Systems, with 90% or more of the respondents stating that they
would like to use some of the services. Approximately 60% responded that they
would particularly like to use the support for prevention of crossing collisions
and support for prevention of collisions with forward obstacles. This suggests
a high level of expectation with regard to these services.
When asked what factor they considered most important for practical application
of Cruise Assist Systems, 51% cited low cost and 60% cited system reliability.
The fact that a majority of respondents emphasized such practical considerations
for the dissemination of Cruise Assist Systems was a distinctive result of the
survey. (Fig. 14.)
| Fig. 14. Results of Questionnaire Survey at Public Tests |
 |
4. Plans for Future Testing
The primary purpose of the proving tests conducted in 2000 was to confirm the
practical feasibility of Cruise Assist Systems that coordinate the AHS and ASV.
These tests involved verification of the validity of the infrastructure system
design figures (verification of the requirements and the systems), of the effectiveness
of services, and of driver acceptance. However, the tests were carried out under
the limited conditions available on the test course. As a result, they did not
adequately test the systems in a variety of road shapes, and did not adequately
verify them in conditions closer to the actual traffic environment, such as
multiple vehicles, multiple lanes, and simultaneous provision of combined services.
In the course of these proving tests on the test course, other issues were also
brought to light that will have to be dealt with in the future. It will be necessary
to create new plans for the future that account for the verification of these
matters.
4.1 Issues that Remain from the Proving Tests in 2000
| (1) |
Testing on varied road shapes and in diverse traffic environments The recent proving tests were implemented only on the road shapes and within the range of conditions available on the test course. Consequently, it will be necessary to carry out tests on more varied road shapes. Furthermore, the range of the testing was also limited in that it did not go beyond the scenario of services provided to single test vehicle in a single lane. It will be necessary to implement testing and verification in an expanded range of conditions that includes more ordinary traffic, typical lane configurations, simultaneous provision of multiple services, and mixed traffic. |
| (2) |
Tests Using Spot Communications In April 2001, the Ministry of Public Management, Home Affairs, Posts and Telecommunications issued revised Radio Law enforcement regulations that relate to DSRC, and AHS services will utilize spot communications for the time being in order to conform with these regulations. The proving tests carried out in 2000 used continuous communications, as called for in the requirements, so it will be necessary to add tests using spot communications while retaining the lessons learned from that previous testing. |
| (3) |
Tests of Services in Snow Cover and Frozen Conditions Verification of service effectiveness and verification of element technologies require testing performed under a variety of different weather conditions. It will be necessary, therefore, to conduct further tests of services in cold districts and under snow cover and frozen conditions. |
| (4) |
Verification of Fail-safe Functions and Driver Acceptance Tests have not yet been conducted from the infrastructure perspective in order to determine, for example, how infrastructure is to be involved in fail-safe functions by investigating how those fail-safe functions affect the driver. Tests of this kind must be incorporated in the future program. |
4.2 Policy in 2001 and Beyond
It will be necessary to conduct proving tests designed with the following points
in mind in order to proceed toward the objective of practical application while
also dealing with issues that remain from 2000 in the testing and verification
program for 2001 and beyond.
| (1) |
Approach to the selection of test locations The approach to selection of test locations is based on the assumption that tests involving test course scenarios should be separated from tests in actual traffic environments. Under this approach, tests that involve scenarios of dangerous situations and those that involve the recreation of typical situations for verification of various issues are assigned to test courses and to roads that have not yet been opened to the public. Unopened roads will be searched for road shapes suitable for tests that could not be performed on the test course, in particular. Roads that are open to the public, on the other hand, will provide actual traffic environments. These environments will be important in testing for items that were not anticipated during design and for the emergence of any problems. The test course itself will probably need to be refurbished and modified for these tests. It will also be necessary to proceed with study to determine the types of actual roads that should be selected for testing. Such study is currently underway. However, we hope to proceed by selecting roads that satisfy the necessary conditions for testing particular services, and then to implement the testing accordingly. For example, some tests involve combinations of multiple services, such as lane departure and road surface information, and these will require long sections of roadway, double lanes, and sudden localized changes in road surface. We intend to search for areas where long sections of roadway include straight sections, curves, and slopes for such tests. (Fig. 15.) |
| Fig. 15. Approach to Selection of Test Locations |
 |
| (2) | Candidate test locations (under consideration) Study is underway of test locations where both expressways and ordinary roads can be used for every service. Our intention is to implement tests on these kinds of road while distributing the particular functions of those tests among roads that are open to the public, roads that have not yet been opened, and the test course, respectively. |
| (3) | Master plan for 2001 and beyond At this point, the overall schedule for 2001 and beyond includes AHS testing on the test course and actual roads that will be conducted in the second half of 2001. Subsequent testing is slated to consist of joint tests with the ASV, likewise on the test course and actual roads, starting in August 2002. We also plan to proceed as far as possible with work on prerequisite conditions for testing, definitions of services, and construction of other infrastructure. (Fig. 16.) |
| Fig. 16. Master Plan for 2001 and Beyond |
 |
5. Conclusions
To summarize, one conclusion that can be drawn regarding the proving tests in
2000 is that they demonstrated the possibility of constructing a real-time cooperative
vehicle-highway systems. Another such conclusion is that the test verified the
effectiveness of system services and, despite various limiting conditions, they
demonstrated that the many safety support services provided by AHS are effective
in actual accident prevention. Furthermore, the tests provided a means of identifying
issues related to the upcoming transition to practical application. Finally,
the major manufacturers that are involved on both the vehicle and infrastructure
sides were united in a common effort toward practical application by means of
these tests. In this way, a foundation was laid for further studies that will
be directed toward that objective.
A number of issues still remain to be resolved by tests in 2001 and beyond.
We intend to pursue these matters in order to work toward practical application
of these systems, to follow through on resolution of issues, and to continue
the program of verification using the test course and field operation tests
in combination.
|