 |
| |
 |
Akio Hosaka,
Advanced Cruise-Assist Highway System Research Association (AHSRA) |
|
| |
1. Background of Research Related to Joint System Definitions
Formulation of concepts and development of phase 0 and primary requirements for Advanced Cruise-Assist Highway Systems (AHS) began in fiscal 1996. Meanwhile, the Advanced Safety Vehicle (ASV) project has proceeded with development of autonomous vehicle systems and cooperative vehicle-highway systems through its second-term research and development activities.
In fiscal 1999, we defined proving test systems for the eight systems that we refer to as the seven safety services of cooperative vehicle-highway. In fiscal 2000, we conducted proving tests of those systems as defined.
We have put the conceptual approach on the AHS side into some systematic order by combining the requirements for practical application with the results from the proving tests of fiscal 2000. On the ASV side, the development guidelines have been established through the second-term ASV research and development activities. The product of our joint effort to bring together our respective findings and determine what kind of system we will actually create from this point forward is the present joint system definitions.
In fiscal 2002, the joint system definitions we have defined will be built as systems for proving tests, including testing on actual roads, and those tests will be carried out. Then the system definitions will be improved as necessary. This is the cycle we expect to follow. |
| |
2. Situating the Joint System Definitions in Context
There are portions of these systems that ASV and AHS will implement jointly, and other portions that each will implement independently. The present definitions apply to those cooperative vehicle-highway systems that ASV and AHS will implement jointly. (Figure 1) |
| |
 |
| Figure 1 |
|
| |
| The joint system definitions were created in order to clarify the quantitative requirements and functions that must be mutually provided, at a minimum, in those parts of the system that are common to both infrastructure and vehicles and in those items that affect both sides. These were considered mainly in terms of their technical feasibility, and it will be necessary to give separate consideration to the aspects of their cost, the constraints on their actual deployment, and other such factors, as well as to their feasibility in a social sense. |
| |
3. The Flow of Research on the Joint System Definitions
The system intended for the joint tests implemented in fiscal 2000 on the test course of the National Institute for Land and Infrastructure Management (NILIM) was defined and then the results of those proving tests and so on were incorporated into these definitions. Up to that time, it was assumed that the system would employ continuous communications. In the process of practical application, however, the spot telecommunication system was adopted instead, and this change is reflected in the present document. (Figure 2) |
| |
 |
| Figure 2 |
|
| |
The survey and analysis of traffic accident data included detailed analysis of accidents occurring during right turns at intersections. Intersection behavior and so on of motorcycles was subjected to a behavioral survey using video and other such materials.
The schedule calls for plans to be made jointly with ASV for the fiscal 2002 proving tests, and for tests to be conducted during fiscal 2002 at the NILIM test course and on actual roads. The results of these tests will then be incorporated and the AHS system requirements will be defined more clearly. This research flow is intended to allow these activities to be reflected in early practical application of safety services. |
| |
4. Basic Conceptual Approach to System Construction
4.1 Road-Vehicle Allocation of Information Collection
The vehicle side will be allocated the visual range that can be covered by the driver's vision and autonomous vehicle systems. The road infrastructure will be allocated provision of the range of information that vehicles cannot easily acquire, such as information on obstacles ahead on curves and at intersections, road surface information, road shape information, and so on. This approach basically allocates to vehicles coverage of the range that is visible from vehicles, while the infrastructure is allocated support for locations that are not visible or not easily visible from vehicles.
4.2 Providing Information to Drivers
Information from the infrastructure is utilized on the vehicle side, where judgement is made about the content and timing of information provision, and the content of support is determined.
4.3 Avoidance of Danger
The level of support envisioned for the present system is based on the provision of information included under AHS-i that is intended to make drivers be careful. The level of support realized by each service will be use by the vehicle of information that is provided from road infrastructure under AHS-i, and implementation of support on the vehicle side. This does not negate any use of information by the vehicle for warning or operational support, but is rather intended to freely present the vehicle with information for any appropriate use. Naturally, the concept that ultimate judgement and responsibility belong to the driver remains unchanged from before. |
| |
5. Overview of Joint System Definitions Revisions
A systematic comparison has been made of the definitions from the time of the fiscal 2000 joint tests and the definitions for the fiscal 2002 proving tests. (Figure 3) |
 |
| Figure 3 |
|
| |
| The services considered here numbered seven services in fiscal 20002. However, the prevention of crossing collisions includes two systems, one for support of vehicles approaching an intersection and one for support of vehicles stopping and restarting into an intersection. Therefore a total of seven services and eight systems have been defined. It has been determined that support for prevention of collisions with forward obstacles cannot be realized using spot communication, so the formulation of its definition has been postponed. It has been determined to be possible, however, to use information on stationary and slow-moving vehicles ahead provided on a spot basis, and new definitions have been formulated separately. (Figure 4) |
| |
 |
| Figure 4 |
|
| |
6. Fiscal 2000 Joint Test Results and Reflection
The driver reaction times to the information provided by the various services were measured from AHS test data and ASV empirical values. (Figure 5) |
| |
 |
| Figure 5 |
|
| |
| These are the results from deceleration testing. (Figure 6) |
| |
 |
| Figure 6 |
|
| |
The question is how much deceleration the driver applies, for example, when information is received about an obstacle on an expressway. These test results show figures from 1.8 m/s2 to 2.6 m/s2, for an average value of about 2 m/s2, or approximately 0.2 G. A deceleration of about 0.3 G was assumed initially, but the average test results indicate 0.2 G. Here, therefore, the design value was set at 2 m/s2, which agrees with the ASV.
Heavy duty vehicles cannot achieve such rapid deceleration for a variety of reasons, such as the weight of the load being carried, the presence of standing passengers, and so on. For the time being, therefore, the design value is set at 1.0 m/s2. It has been decided to judge the validity of these values through tests and so on conducted in the future. |
| |
7. Examples of Survey and Analysis of Traffic Accident Data
7.1 Analysis of Intersection Right-Turn Accident Patterns
Materials published by the Institute for Traffic Accident Research and Data Analysis (ITARDA) include micro-data from extremely detailed analysis of intersection accident causes. Data of this kind from a period of several recent years was extracted and the patterns in 37 right-turn accident cases were analyzed. (Figure 7) |
| |
 |
| Figure 7 |
|
| |
There are cases when, despite the presence of an oncoming vehicle, the driver at least does not see it, or the driver is not aware of having seen it; and there are cases when, despite having seen the oncoming vehicle, the driver makes an error in judgement. We on the infrastructure side provide support for the former cases.
It is known that the accident patterns that are important to detect involve objects that are not visible because they are blocked by an oncoming right-turn vehicle, not visible because they are blocked by the lead oncoming vehicle, or not visible because they are blocked by a congested line of oncoming vehicles. |
| |
7.2 Overview of Behavioral Survey of Motorcycles at Intersections
It is known that an extremely large number of accidents at intersections result in damage to motorcycles. The detection of motorcycles, therefore, is crucial.
The conditions at accident-prone intersections in the Kanto Region were recorded on video and analyzed. Based on the results, four patterns were identified that present the possibility of conflict during right turns at intersections. It was found necessary to take measures on the assumption that the speed of motorcycles squeezing through traffic during congestion and so on will be 25 - 30 km/h, since the average speed in such cases is 22 km/h, and the 90th percentile is 27 km/h.
An explanation of the specific patterns will follow. (Figure 8, Figure 9)
|
| |
 |
| Figure 8 |
|
| |
 |
| Figure 9 |
|
| |
|
Pattern 1: Squeezing through when the way ahead is blocked by congestion
This is the situation when the way ahead in the intersection is blocked by congestion. The vehicles following behind stop and wait instead of entering the intersection, so vehicles turning right can make their turns. This is when there is danger of collision with a motorcycle that is squeezing through the stopped traffic.
Pattern 2: Squeezing through during ordinary traffic movement
Here the way ahead is not fully blocked, and traffic is moving at some speed. A motorcycle or other vehicle squeezing through at high speed may conceivably collide with a vehicle making a right turn.
Pattern 3: Squeezing through when there is a vehicle decelerating or stationary because it is making a right or left turn
In this scenario, the lead vehicle decelerates in order to make a left turn. Then a vehicle comes speeding up from behind when it seems possible for it to make a right turn as a result, or a motorcycle that was beside the left-turning vehicle squeezes through from behind it.
Pattern 4: Squeezing through when there is a vehicle decelerating or stationary because it is yielding the way to an oncoming vehicle or the light has changed
This is a so-called "thank-you accident," when a vehicle stops to yield the way to a vehicle making a right turn, and a vehicle squeezing through alongside the stopped vehicle collides with it.
|
| |
7.3 Design Speed for Infrastructure Disposition
We determined to execute the safety design on the condition that our design speed would cover speeds up to about the 90th percentile of speeds recognized as dangerous in accidents that resulted in death or injury.
The speed recognized as dangerous is the speed at which the driver causing an accident realizes the danger and immediately before taking evasive action. Cost factors, site limitations, and so on make it infeasible to cover all speeds. We determined, therefore, to work on the assumption of about 90% coverage. This corresponds generally to the speed limit plus 30 km/h. |
| |
8. System Specifications
8.1 Start Information DSRC and Information DSRC
When providing information within a certain zone, there will be a certain amount of radio wave leakage into the surrounding area, no matter how well adjusted the antennas are for high performance. This means that when the system tries to transmit information about a problem ahead in one lane, a vehicle in the opposing lane may end up receiving the information, confusing the driver when there actually is no problem. It is necessary first of all, therefore, to notify vehicles of which information is valid for vehicles traveling in which direction. For this purpose, information is to be considered valid when it is received after passing a start DSRC. (Figure 10) |
| |
 |
| Figure 10 |
|
8.2 Relative Locations of Incident and Information Transmission
This formula is used to derive the distance L from the changes in speed, including driver reaction time, when the vehicle decelerates. (Figure 11) |
| |
 |
| Figure 11 |
|
| |
8.3 Extending the Service Target Zone Distance and Service Zone Length
It is possible for a single information DSRC to transmit information from a considerable distance ahead. Information on the road surface or on a curve, for example, or other such information that does not change greatly over time, can be provided rather far upstream and be kept valid for a considerable time. In such cases, a greater distance can be covered without placing two sets of beacons if a position compensation DSRC is placed about every kilometer. The idea is that the vehicle can use this to compensate for its own cumulative error. (Figure 12) |
| |
 |
| Figure 12 |
|
| |
8.4 Information Provision and Driver Reaction Time
The driver reaction time is not determined simply by adding together the information provision time and the reaction time. It is also assumed that drivers may actually begin to react at a certain point after information has begun to be provided. Therefore we assume that drivers will take action after about 5 seconds in a road section of uninterrupted flow field, and about 4 seconds in an intersection system. (Figure 13) |
| |
 |
| Figure 13 |
|
| |
9. Example of System Configuration
9.1 Support for Prevention of Crossing Collisions (Support When Stopping and Restarting into Intersection)
A vehicle moving on a smaller road will stop at an intersection, and if visibility is poor because of corners and so on, it will head out into the intersection in order to check for vehicles moving on the through road. This service seeks to assist by providing information on whether there is danger at an extremely close range. (Figure 14) |
| |
 |
| Figure 14 |
|
| |
| The sensors are arranged on the assumption that they should detect motorcycles squeezing through at the same time they detect automobiles and so on. Some accidents occur when a bicycle traveling along the left shoulder of the road collides with a vehicle. Therefore bicycles and other such objects nearby on the road shoulder are also included as detection targets. |
| |
9.2 Support Service for Prevention of Right Turn Collisions
This service detects vehicles that are squeezing through and so on, including vehicles moving along between the road shoulder and the traffic lane. When the intention is to convey information as early as possible to a vehicle approaching at a fairly slow speed and moving into position for a right turn, it is necessary to provide information somewhat farther before the intersection. The service is designed to convey information with two types of DSRC in order to cover this eventuality. (Figure 15) |
| |
 |
| Figure 15 |
|
| |
 |
| |
