AHS development has three general objectives. The first is to resolve the various issues of road traffic by providing new traffic systems or traffic support systems that improve the safety and efficiency of road traffic by giving roads advanced detection and communications functionality that enable vehicles to make use of the information acquired.

  The second is to establish, by means of AHS development, the basic sensing and communications technologies that are crucial to the construction of a common platform for ITS applications. This will allow work on the various ITS applications that are related to roads to be carried out so as to coordinate those applications efficiently and organically with each other.

  The third objective is to contribute to more advanced and more efficient road management by making multipurpose use of developed AHS technology. These are the three general objectives we have set ourselves and given thought to.

  As to the concept, the first point is the necessity for countermeasures immediately before an incident. Figure 2 shows a breakdown (as of fiscal 2000) of accident causes.

Figure 2

  A variety of different measures have been implemented for traffic safety in the past. These have ranged from traffic regulations and restrictions, driver education, pedestrian education, improvement of safety facilities, and other such measures taken in advance, to improvement of automobile body construction, use of airbags, and other such measures to reduce damage immediately after accidents occur. These have reduced the number of traffic accident fatalities considerably, but the number of traffic accidents continues to remain at a high level. Analysis of the causes of those traffic accidents shows that 75% result from human error in recognition, judgement, and operation. The importance of measures to support driver avoidance of accidents immediately before they occur is clear.

  Another point here is cooperative vehicle-highway. Progress is being made in improving vehicles for safety and so on. However, there are areas that vehicles alone cannot deal with adequately, such as acquiring information on blind spots and areas far ahead, obtaining a variety of different perspectives on events, making comprehensive judgements based on consideration of overall traffic or the region as a whole, and so on. It goes without saying that the infrastructure side also has its limitations. Both sides have their respective strong points and weak points. (Figure 3)

Figure 3

  By taking advantage of the strong points on both sides and complementing respective weak points, we become able to support safe driving across a fairly broad range. Cooperative vehicle-highway has come to be recognized as extremely important for the future. (Figure 4)

Figure 4

  Japan has been a leader in this field, although America has recently understood that there are limits to what vehicles alone can accomplish as countermeasures in intersections and so on. They are carrying on development of information provision using infrastructure sensing and communication to provide information to vehicles, and these methods have been demonstrated in action.

  As I understand it, therefore, the importance of cooperative vehicle-highway has come to be recognized not only in Japan but also in other countries. That recognition underlies our basic conceptual approach of cruise assistance through cooperative vehicle-highway mediated by communications between road infrastructure and vehicles.

  We have studied measures to achieve realistic progress toward actual evolution of the concept.

Figure 5

  Figure 5 shows the results. We worked out the concept of AHS-i, -c, and Ða support levels. The i, for information, is information provision and information assistance; the c, for control, is operational support; and the a stands for automated cruise.

  The relationship between information assistance and operational support at these levels is shown in the figure. Another major topic here, however, is the problem of responsibility. The basic assumption here is that when we come to the a automated cruise level, all responsibility is borne by the system. Initially, quite a number of studies were made of the possibilities of automated cruising, but for practical reasons, we have since decided to work from information provision and operational support. Conceptually, therefore, we have worked out an approach through AHS-i and Ðc.

  There are various conceivable ways to approach this evolution (Figure 6).

Figure 6

  For example, one approach (b) would be to expand the scope of system assistance from i to c and a, and then the scope of use could be expanded after that. Alternatively, the scope of use could be expanded from the start, even if the level of assistance remains low, and the scope of assistance could be expanded after achieving a certain degree of social acceptance (a). We have adopted the latter approach (a), and have recommended this as a realistic way of promoting AHS.

  The content of cruise assistance was analyzed in order to systematize the fundamental user services (Figure 7).

Figure 7

  Our analysis, for example, took a time axis, according to whether countermeasures were immediately before accidents or in advance of accidents. We also analyzed in terms of goals, such as problems of safety, efficiency and the environment, and smoother traffic flow, and also in terms of the behavior of the driver being assisted, such as longitudinal behavior, lateral behavior, and intersection behavior.

  We decided to address the fundamental user services by first dealing with problems of safety. We performed analyses of accident statistics and accident causes. Figure 8 shows one example of this. The colored portions to the right represent delayed recognition, the portions to the right of that represent errors in judgement, and to the right of that are errors in operation. This suggests the extent of impact to be expected from different levels of assistance. The lengthwise width of these bands indicates the magnitude of loss and the number of accidents, showing the extent of coverage to be expected by separate types of assistance.

Figure 8

  This type of analysis was carried out basically for ordinary roads, then for expressways, and then overall. We determined that, of the fundamental user services, the most important are services for Maintenance of safe headway, Prevention of collision with obstacles, Lane keeping (straight lane), Lane keeping (curves), Prevention of crossing collisions, Prevention of right-turn collisions, and Prevention of collisions with pedestrians crossing streets, and we decided to give priority to development of these services. Drivers perform recognition, judgement, and operation, and we worked out the concept of assisting those by information provision (assisting recognition), warnings (assisting judgement), and operational support. These three are to be implemented by cooperative vehicle-highway (Figure 9).

Figure 9

  We thought through the requirements for achieving these in practice.

Figure 10

  Figure 10 presents a conceptual image of assistance. The flow of services is shown in the vertical direction. First, information is provided concerning an obstacle. If the driver receives that information and responds appropriately by decelerating and stopping, the danger will be removed. If this does not happen, and, for example, the driver continues forward without reducing speed, then a warning is given. If for some reason safety measures are still not taken after that, then operational support is implemented by automatic braking and so on.

  Certain fundamental principles will govern whether the services produce their full impact. It is crucial for the services to be widely accepted, that the great majority of users accept them comfortably, and that the services can be implemented at appropriate levels of technology and cost. We formulated the requirements and have been refining them with these principles in mind. (Figure 11)

Figure 11

  Figure 12 shows one part of the contents of AHS requirements. Take the scenario of a vehicle, for example, that approaches at a certain speed, discovers an obstacle, and comes to a stop. If the vehicle can stop before the obstacle, then an accident is averted. The process up to stopping involves driver reaction time, the time required to decelerate, and so on. Considering these factors, then the amount of time in advance that warnings have to be given, and the amount of time in advance that information should be provided, are values determined by speed and so on. Statistical data on accidents can give a fairly clear idea of how fast accident vehicles were traveling. We add something of an extra margin to that speed, and then begin to derive increasingly specific requirements that can be applied to that range of speeds.

Figure 12

  We created a basic system on the test course and subjected it to tests. Various points emerged that required us to devise solutions for actual use. Initially, for example, we assumed the use of continuous communications, but we then discovered that it was more realistic to begin with a spot communication system, in which transmissions are received at a single location. Taking technical limitations and other factors of this kind into consideration, we consulted with the ASV project and created a joint system definition, then set realistic development targets directed toward proving tests of actual working systems.

Figure 13

  Figure 13 presents the example of Support for Prevention of Collisions with Forward Obstacles (support for detecting forward standing and slow vehicles). If information provision is to take place, warnings issued, and brakes applied, then during that process it will be necessary to provide information continuously to the vehicle in real time.

  Realistically, however, the technology we are applying will provide information at a single location. This means that a continuous flow of information provision of the kind we had envisioned initially will not be possible. In this case, therefore, the system will actually be providing information on the order of that on an information board inside the vehicle. This has the advantages that we can make the information easier for the driver to understand, and handle it to take account of the vehicle's speed. In any event, we came to realize that we can only offer a service very much like that of an information board, and we took that into consideration in setting the requirements.

  Next we set targets for safety and reliability, which are performance objectives on the infrastructure side. First of all, these words "safety" and "reliability" are used in many different ways by different people. In order to define them properly, we created clear descriptions of 14 states of safety and reliability (Figure 14).

Figure 14

  We defined these states first of all in terms of whether they involve danger or not, and then whether they will be conveyed to the driver or not.

  As a result, the first area has an event that is actually occurring and that is accurately conveyed to the driver, or even if there is a malfunction, the fact of a malfunction will be made known to the driver, so there is no problem. The second area has a situation of actual danger but this is not conveyed to the driver, or there is a system malfunction that is not conveyed to the driver. Therefore this is a dangerous condition. The third area has various causes for concern, but nothing that directly entails danger. Our approach was to divide events into these three areas and attempt to reduce the dangerous conditions (marked with a Z in the figure) as much as possible.

  The conceptual approach we followed in setting actual targets was as follows:

  First, we defined cruise-assist system safety and reliability to conform with JIS.

  In terms of needs (the user side), there is a requirement to set the safety target as high as possible. We intend to respond to this need as far as is realistically possible, and will make every effort in that direction.

  In terms of the existing technical and economic resources for infrastructure to realize that target, however, it is not possible to achieve 100% safety or reliability (which you can also think of as availability rate). Consequently, we decided to set safety and reliability targets that are realistically possible to achieve, and intend to go on from there to address any issues that arise.

  Taking a comprehensive view of the needs and resources into consideration, we decided to use the JIS safety integrity level (JIS B508 defines the concept of the safety integrity level for systems of this kind, and sets target values) for the proving test systems. Our objective was to assure a comprehensive safety of 90% or better as prescribed for safety integrity level 1, for which we set the provisional safety target of 95% or better, and the service availability rate of 95% or better.

  Even if these targets were achieved, there would still be a problem from that portion which actually fell short of 100%. The remaining 5% or 10% that did not meet the target must not become the location of new danger or of otherwise extremely serious circumstances. We agreed, therefore, to implement failsafe measures extending to the vehicle and human systems.

  The targets for safety and reliability were set according to this approach. We devised supplementary (failsafe) safety and reliability measures for vehicles and for infrastructure systems development. We then verified our hypotheses using tests on a driving simulator, test course, and actual roads, and intend to apply them to actual working systems. These are the objectives for the requirements we formulated.

  The failsafe measures I just mentioned were then examined in terms of the 14 states shown in Figure 14. We studied the measures to determine what kind of display in the vehicle for each would avoid danger, and also, if those problem states did occur, whether they could be prevented from tipping into a negative direction. As a result, we devised a three-state display. (Figure 15)

Figure 15

  In the event of actual danger, the driver is naturally notified of that danger. Even when there is no danger, if the vehicle is in the kind of location we are thinking about, such as a dangerous intersection or curve, then the system will display a message to the effect that this location is often dangerous, so the driver should be cautious. This is done in order to keep drivers from allowing the absence of information to lull them into taking some action that is dangerous. When the system stops operating or malfunctions, it will stop providing information. In that case, messages to promote caution and information to encourage careful driving will also stop appearing. The display will be completely blank. When drivers see a completely blank display, they should notice that they are not receiving information as they usually do in that location, and realize that the system is not operating normally. Our idea is that this will prevent reckless driving when the system is malfunctioning or when the system does not provide any information because it is having a hard time detecting conditions. (Figure 16)

Figure 16

  As I said at the beginning, our basic conceptual approach is to assure the highest level of safety possible. Therefore we analyzed the problems that could occur in the system and in the element technologies, analyzed their causes, and implemented countermeasures. These efforts to improve safety to the greatest possible extent have been carried on side by side. (Figure 17)

Figure 17

  The result has been, as shown in Figure 18, that we have been able to set specific targets in order to achieve an overall safety of 90% or higher. The targets are for safety of the separate items of sensor equipment required to achieve that level, the safety of road-to-vehicle communication facilities, availability rates, and so on. We used these target values to press forward with our research and development, and this brought us to field operation tests.

Figure 18