Sub-Working Group on Social Rules for automated driving vehicles in age of AI (Second)

Overview

Date and time: January 30, 2024 (Tue) from 10:00 to 12:00
Location: Online
Agenda:
1. Opening
2. Content
  1. Explanation by the Secretariat (Main opinions of the First age of AI Sub-Working Group on the Review of Social Rules for automated driving vehicles and Possible Issues Based on the Opinions)
  2. Status of architectural studies on automated driving security
  3. Accident investigation and information sharing to determine the cause and prevent recurrence
  4. Data connections and how it should be
3. Exchange of opinions
4. Adjournment

Materials

Minutes

Councilor Kodama: First of all, I would like to make an administrative contact. Today's meeting will be held completely online. Members are requested to turn on the camera during the meeting and to unmute the microphone when they speak. If anyone else is speaking, please mute them. In addition, spectators are requested to turn off the camera and microphone.

Next, I would like to check the materials. As stated in the agenda sent in advance, the materials include the agenda, the list of members, the secretariat explanatory materials, the Information-Technology Promotion Agency Digital Architecture and Design Center, Mr. Ouchi's explanatory materials, the Member Suda's explanatory materials, the Member Nishinari's explanatory materials, the materials to be submitted by the members, and the list of attendees. If you do not have them, please contact the Teams chat function or the secretariat by email. As for the attendees today, due to time constraints, I would like to introduce them by distributing the list of attendees you have. In addition, I heard that Member Harada is scheduled to leave the meeting in the middle of the meeting. Please be aware that the materials and minutes of the meeting will be made public at a later date.

Then, I would like to ask Mr. Kozuka to proceed from here. Mr. Kozuka, please.

First, I would like to explain the secretariat of Agenda 2 - (1). I would like to hear an explanation of the main opinions of the first Sub-Working Group and the expected issues based on the opinions. Mr. Suga, Counselor of the Secretariat, nice to meet you.

Counselor Suga:
(Hereinafter, explanation will be made based on "Material 3: Second Secretariat Explanatory Material")
Material 3: Based on the 2nd Secretariat Explanatory Material, we explained the main opinions at the 1st SWG and expected issues based on the opinions at the 1st SWG.

Mr. OUCHI: In terms of how to collect the first
(The following explanation is based on "Material 4: Status of Architecture Study on automated driving Security")

I am Ouchi from the Digital Architecture and Design Center of the Information-Technology Promotion Agency. I am pleased to meet you today. I would like to report on the status of architectural studies on automated driving security. Please take a look at the next page.

First of all, regarding today's premise, the purpose is to improve automated driving vehicles security by collecting and sharing data on automated driving vehicles accidents and close calls to the extent necessary. Future discussions will include how to handle data on accidents and close calls and how to handle all cases. In addition, the use of data is limited to the investigation of causes, prevention of recurrence, and prevention of accidents. Regarding collected data, we have determined the use of data and are considering it on the premise that it will not be used for the purpose of pursuing legal responsibility. It will be necessary to discuss who will collect and share data and in what form. In addition, I would like to introduce overseas cases as reference information today. Please take a look at the next page.

Here, we show the policies for the Comprehensive National Development Plan for Digital Lifelines. DADC is considering an architecture to respond to people's flow crises, distribution crises, and increasingly severe disasters. The architecture includes hardware, software, and rules, and we are inviting business operators to DADC to consider the architecture. It has already been announced that automated driving vehicles lanes for implementation have been set in Suruga Bay Numazu, Hamamatsu, and other places since fiscal 2024. Based on the implementation in such places, DADC is considering developing the architecture over a 10-year period toward mid-to-long-term social implementation while continuing to brush up on the architecture. Please go to the next page.

In cooperation with the Ministry of Economy, Trade and Industry and Digital Agency, DADC serves as the secretariat of the National Comprehensive Development Plan Conference on Digital Lifeline Infrastructure, which is subordinate to Conference on the Realization of the Digital Garden City Nation Vision and Council for the Promotion of a Digital Society. Under the National Comprehensive Development Realization Conference, five working groups are composed of three working groups related to the Early Harvest Project, automated driving Support Road Working Group, drones Route Working Group, and Infrastructure Management DX Working Group, as well as Architecture Working Group and Startup Working Group. I think the automated driving Support Road Working Group is the target of today's introduction. Please go to the next page.

In the materials of the 2nd automated driving Support Road Working Group of Digital Lifeline, this article extracts the parts related to automated driving safety. In the 2nd Sub-Working Group, we show that in order to respond to risks that are unknown in location and response, it is possible to collect data by connecting systems of each company and platforms developed according to the purpose, and to prevent accidents by performing simulations in a virtual space based on the collected data. In addition, the identification of the causes of accidents and the pursuit of responsibility are outside the scope of this material. As shown on the right side of the material, in the Early Harvest Project, we have also shown at the 2nd Sub-Working Group that we can reduce risks to automated driving by making the 2nd traffic zone from the Numazu Service Area in Suruga Bay in Shintona to the Hamamatsu Service Area a automated driving vehicles priority lane. Please go to the next page.

Regarding the concept of preventing accidents, we have received comments from business operators that it is necessary to conduct a huge amount of testing driving in order to measure risk scenarios based on near-miss information that occurred during driving at each company in the current Issue, and it takes time to ensure safety and validation. In addition, Issue has raised the possibility that it may take time to identify the cause or that the cause cannot be determined due to a lack of information because all services may have to be stopped until the cause is identified due to the cooperation of various systems when accidents occur. As a solution to DADC, it is suggested that accidents and near-miss incidents can be reproduced in virtual spaces by collecting information and near-miss information at the time of accidents and generating and sharing environmental models, and that accidents can be prevented by validation development vehicles, including those already released to the market, in virtual spaces and updating software as necessary. Please go to the next page.

Here, I would like to show an embodiment of the previous part. In the part on the left side of the document, I would like to suggest that a certain amount of time should be sent at the time of detection. At the time of detection of a close call, based on triggers such as unexpected actions, equipment failures, and detection of acceleration exceeding a certain value, which are not definite, data necessary for generating an environmental model that reproduces accidents and close calls is sent to the information collection organization from operators and non-operators of the automated driving vehicles. We are considering sending the data to the simulation vendor side, so that the simulation vendor can generate an environmental model that can reproduce accidents and close calls, and use it to provide an environmental model to development operators, insurance companies, research institutes, etc. in the automated driving, so that accidents can be reproduced in a virtual space. In addition, I would like to see it used by the public authorities and accident investigation agencies, etc. in the future. Please show me the next page.

This shows an image of generating an environmental model that reproduces accidents and close calls. What an environmental model is like is shown in the picture below on the left. We are currently considering what kind of data is required to create such a model in a virtual space. For example, drive recorder images, sensor information, vehicle speed and acceleration, etc., can be considered. In the future, we will clarify what kind of data is required to be collected and by what means in the Early Harvest Project to be launched in fiscal 2024. In the reference case on the right, the upper half describes the information required to be provided by automated driving operators, which is listed in the German Road Traffic Law and the Chinese Guidelines. In addition, the lower half summarizes the information required to be provided by Chinese operators other than automated driving operators. Based on this information, we will consider what is required to generate an environmental model that reproduces accidents and close calls. That is all from DADC.

Mr. Suda: You are right that although you say it is not a place to pursue responsibility, there is a concern that it will be used to pursue responsibility. As I said last time, I think it would be good to have some kind of system like immunity from liability from a long-term perspective.
(The following explanation is based on "Document 5: Accident Investigation and Information Sharing to Determine the Cause and Prevent Recurrence")

I'm Suda from the University of Tokyo. I have some materials ready for you today, so I'd like to give you a brief explanation. Nice to meet you. Next slide, please.

I myself am an engineer, and I have been conducting research development in various mobility including automated driving. Please go to the next page.

I would like to talk about the accident investigation this time, but before that, I would like to briefly introduce what I have been involved in. As I told you last time, since I joined the investigation and review committee on the derailment of the Hibiya Line in 2000, I have been in charge of various accident investigations. Most recently, there was a case in which a light rail in Utsunomiya derailed during a test run the year before last, and I am investigating the cause and considering countermeasures. In addition, the Aircraft Railway Accident Investigation Commission was established, and it became the Japan Transport Safety Board. As an expert member of the Aircraft Railway Accident Investigation Commission, I participated in the investigation of the derailment of the Joetsu Shinkansen, the Tohoku Shinkansen, and the Kyushu Shinkansen, as well as the derailment of the Fukuchiyama Line. For this reason, I was appointed as the chairman of the recently established automated driving Accident Investigation Commission, conducted an investigation into the collision of a medium-sized bus in the Paralympic athlete village, and issued a report last summer. Please turn to the next page.

In addition to the activities described above, we are engaged in various activities in the field of safety. At SCJ, we are active in the safty and security Risk Review Subcommittee of the Comprehensive Engineering Committee. As a job of the Maintenance Section of the Distribution and Automobile Bureau, OBD vehicle inspection is about to start working, and we are also participating in the Review Committee. I think OBD is one of the ways to use it, but I will omit an explanation today. Other than that, there is the Automobile Products Liability Consultation Center, which is also in charge of examination, and we are participating in the ADR method, which does not require a trial, between victims and manufacturers for various failure information. In relation to railways, it is very important to share accident information, and we have decided to create a safety database at the Railway Technology Promotion Center of the Railway Technology Research Institute, which has a history of about 20 years. We have had the experience of considering various methods from data collection to actual disclosure. Please go to the next page.

Let me briefly introduce the outline of the automated driving vehicles Accident Investigation Commission that I mentioned earlier. It is operated by the Traffic Accident Analysis Center, commonly known as ITARDA, on behalf of the three bureaus of the National Police Agency, the Distribution and Automobile Bureau of the Ministry of Land, Infrastructure, Transport and Tourism, and the Road Bureau. As a general rule, the accidents in automated driving vehicles at Level 3 or higher are targeted. However, the mid-sized bus collision that occurred on August 26, 2021 was officially investigated and announced on September 2 last year. Please go to the next page.

It shows the members who submitted the report. I served as the chairman, but in addition, experts in automobiles, law, and administration are participating. It is surrounded by squares in the document, but there is a proviso that this accident investigation was conducted for the purpose of contributing to the investigation of the cause and the prevention of recurrence, not for the purpose of pursuing responsibility. Please go to the next page.

Next, I would like to briefly introduce the content of the report. The cause of the accident was not one factor but a combination of multiple factors. We concluded that the accident was caused by a wide range of factors, including a driver problem, a traffic guide problem, and a problem in the awareness of safety measures by the people concerned. Therefore, a wide range of information, including not only vehicle data but also various systems and infrastructure, was necessary for the investigation. Please go to the next page.

In response, we are also making recommendations to prevent a recurrence. I will omit the details, but we are making various recommendations such as those related to vehicles and services. At the end of this, it is said that record keeping is important. Since the automated driving was erased, we are incorporating this into our recommendations as being extremely important. Please go to the next page.

I would like to introduce the point that became Issue at that time. This material was taken from the accident investigation report. As the subject of the investigation, we investigated the vehicles, the victims, and other persons concerned, but as it is written in U.S. Seal, we could not obtain consent to hear the opinions of the victims and the guide. Since the Investigation Committee is supposed to conduct the investigation after consent, we do not have the authority to investigate, and we could not obtain detailed information, which is essential, so the unfortunate reality is that it is beyond speculation to some extent. Please take the next page.

I myself was in charge of several accident investigations as an expert member of the Japan Transport Safety Board, so I would like to introduce your story. First of all, the Aircraft Accident Investigation Commission, which is under an international agreement, is currently the Japan Transport Safety Board, a combination of railway accidents and part of the Japan Marine Accident Tribunal. I have been involved in the investigation of railway accidents, but there was originally no such organization for railways. However, beginning with the Shigaraki Kogen Railway incident, the need for an investigation by a public self-reliance organization was recognized, and the Railway Accident Investigation Review Board was established and conducted an investigation on the Hibiya organization. Based on the results, I recognize that it has taken on the form it has today. Since the investigation is based on the Act for Establishment of the Japan Transport Safety Board, the committee members and the investigation are independent of government agencies, and an independent investigation is guaranteed. In addition, as part of the authority for the investigation, hearings can be held with various parties concerned. In addition, there are regulations on what kinds of accidents are subject to the investigation, and the investigation cases are selected in accordance with the rules. Therefore, it is a very fair and neutral mechanism. Based on this experience, I think it is very important to create a mechanism like the Japan Transport Safety Board. Please turn to the next page.

We propose activities that are required in the future. It is very important that the accident investigation is not a place to pursue responsibility, but an investigation to determine the cause of the accident and to take measures to prevent a recurrence. It is also very important to make proposals to reduce the damage, even if it is not possible to prevent it completely. In addition, it is of course necessary to make the report of this investigation public and share it so that it will be useful in preventing a recurrence. However, as I stated earlier, what must be considered in doing so is the formulation of rules for investigations and the collection of information from relevant parties, as well as consideration for the victims and bereaved families. I am also involved in the Fukuchiyama Line accident investigation, and I am very keenly aware of it. Another important aspect is the sharing of accident and incident information. It is important to share various data even for cases that did not become investigation cases. In short, from the perspective of accident situation investigation, I think it is necessary to distinguish between individual serious accident investigation cases and statistical cases in which information on accidents and incidents is widely shared. Critical information differs depending on the project, but basically, data such as vehicle behavior is very important. In this sense, it is important to comprehensively collect not only data recorded only by vehicles but also various factual information such as drive recorders and surveillance cameras. However, in recognizing facts as facts, it is confusing if the time of these information is not correct. That is why I am acutely aware that time alignment is very important. In addition, for projects other than individual investigation projects, as I introduced earlier, it is important to create a mechanism to collect a wide range of incidents and accidents, such as the Railway Safety Database. Currently, ITARDA is working on a contract basis, but I think it is possible to think of a mechanism to widely share and disclose information on accidents, incidents, and safety information by cooperating with such organizations or by utilizing the mobility Innovation Alliance, a general incorporated association established to continue the results of cross-agency efforts on automated driving during the first and second phases of SIP. That's all from me.

Nishinari Member: Thank you for your question. I think it is a question about German law, as you understand. I am not a legal specialist, so I had a very hard time reading this. Looking at (5), there is no regulation on how to share it, but it says that for accident prevention and scientific research, it can be shared for research purposes or public interest purposes at universities, research institutes, local government, etc., and I think it is a very big step forward. I have not heard about the mechanism, etc., and I think we will discuss it from now on because it was a new story that was completed last year or the year before last.
(Explanation based on "Material 6: data connections and its Future")

Best regards I'm Nishinari. I'm not only studying cars, but also distribution and people's flow, and recently, I'm also studying the safety of flying cars. I have no time today, so I have summarized what I want to say on page 1. First of all, it is the same in any industry, but manufacturers and related organizations independently hold data related to safety, and it is not shared. Therefore, it should be shared by researchers or specialized organizations by skillfully extracting the personal data. I think the information on security is from cooperative areas, not competitive areas, but I don't think the line is drawn properly. This is an actual example of a problem with data that should be retained, but when I tried to investigate it, there were problems such as a lack of data, a lack of essential data due to a lack of standard specifications, a lack of data, and low quality. I think that these should be prepared. In addition, if there is no platform, it takes time to communicate. Therefore, I think it is necessary to create a platform with a secure mechanism with high safety. This was related to SIP in logistics, so I will talk about this later. In addition, I think it is necessary to share information on close call cases. In addition, it is a little different story, but I am also interested in the story of social acceptance, and automated driving vehicles is a means, not an end. If you don't share what it is for, you won't gain people's sympathy. I told you before about why the Apollo program was able to continue despite the deaths. As it is written in a book called Mission Economy, it was because dreams were shared that we could move forward. Next page, please.

As DADC mentioned earlier, I think it is necessary to discuss what kind of data should be prepared in the future. However, I thought about what I would need if I were to analyze it, and dropped it into the document on page 2. On the premise that the personal data has been surpassed, what is required is position, speed, acceleration, and a change in acceleration called jerk is also important in automobiles. In addition, the condition of the car, communication status, external conditions, etc. are also required. You mentioned the case of China, but this research is such a rapidly advancing area that a considerable number of papers are published in a week. It was a malicious interruption that I read the paper last week and couldn't add it to this material in time. It is said that the information I mentioned earlier is not enough when someone acts maliciously and an accident occurs because of it. According to a paper published last week, if a laser beam is applied to a road sign for a moment, a car can make a recognition error and create a situation in which, for example, the car stops beyond the stop line. It is not only a problem of the car, but also what to do when there is a third party with malicious intent, and some of them may be caught by the drive recorder, but I think that the accident will not disappear unless we consider such points. I will introduce it later, but what is not specified even in the German regulations is the data on how other cars behaved. If someone interrupts, we cannot analyze without data. Therefore, it is easier for me to analyze if I have not only the information of my car but also the data of how other cars moved or how people moved. In addition, there are papers on how people move almost every week, and there is a technology called TrajectoryPrediction, which predicts the movement of cars and people around you. The car moves by predicting in a short time where the people and cars around it will be in five seconds, but the algorithm of automated driving vehicles differs from researcher to researcher. There are papers that compare the differences, but for example, the technology for predicting movement is different, such as the prediction of the movement of people and cars by Company A is different from that of Company B. This is a story from competitive areas, but I think that by sharing this information, we can improve ourselves by friendly competition. Another thing, which is also discussed in flying cars, is how to avoid collisions safely, and there is a famous standard called RSS for this. The standard that the other party is bad and it is not your fault because you keep this much distance between cars is proven by a mathematical formula. Everyone uses this formula carefully, but in terms of research, if you use this standard, the distance between cars will be too large, and traffic capacity will be reduced. There is a trade-off that if you prioritize safety, you will lose efficiency, and I think it is quite difficult to run with RSS. Next page, please.

As I mentioned earlier, SIP has been operating a logistics platform for two years. The key point is the non-falsification of data. Since there is a possibility that someone will change the data, there are problems such as how to ensure the non-falsification of data and the control of access authority. In addition, the data formats are different between each company, and when data is collected, it cannot be collected in a word. It takes considerable practical technology to convert, extract, and summarize it. In addition, it is not a closed platform, but it is necessary to cooperate with other platforms in order to match various things such as weather. The rest is data standardization, which we have been working on in the logistics domain for two years. Please come to the next page.

There is a Herbert William Heinrich that there are a huge number of incidents in the background of one incident. I belong to the Department of Aerospace Engineering, and in the aircraft industry, there is a system called VOICE, and I think it is a good system. As Dr. Suda mentioned earlier, a third party organization is contracted to ensure that individuals and company names are not identified, and administrative penalties are not imposed, so it is easy to provide information on close calls. As far as I have researched, 60 to 100 cases a year have been listed. An accident occurred at Haneda the other day, but such runway incursions are not rare, and if you look at the report, you can see many stories of runway incursions. In VOICE, information is collected in a form that does not specify who is responsible. Please go to the next page.

With regard to data, I mentioned an example from Germany earlier, but there is a 1g rule, which is described in detail in the reference materials below. In Germany, the purpose of use, such as the collection and disclosure of data, is specified in the Road Traffic Law. Normally, manufacturers write it down in their own instruction manuals, etc., but it has been pointed out that civil lawsuits are likely to occur if they do so. The German Road Traffic Law 1g is unique and if it is the latest one, in particular, as described at the bottom of the materials, accident victims can also request data disclosure. As Dr. Suda said earlier, various considerations are necessary, and I think we will discuss what to do in Japan in the future, including such considerations.

Below, I would like you to take a look at it as a reference material. On page 6, information that must be collected in Germany is written in red. On page 7, it is written that universities and research institutes can use it, in terms of who can use it and under what circumstances. On page 8, (7) is a new addition, but it is written about the rights that can be claimed by third parties. On page 9, it is written about cases in China. As DADC mentioned earlier, it is interesting that the environment and response status are required as the data to be provided, and these are currently being discussed. That is all.

Member of Hatano: Thank you for your nomination of . First of all, thank you very much to everyone who explained today. In addition, I would like to thank the members of the Secretariat for preparing a very careful exchange of opinions and so on for the second meeting. From the Japan Automobile Manufacturers Association, I will convey my opinions limited to today's issues. There are three points for each issue and one point for the whole.

First, I would like to comment on the mechanism of the accident investigation. The Japan Automobile Manufacturers Association (JAMA) supports the study of a system for accident analysis and accident investigation to prevent recurrence, and a mechanism for handling data in the event of an accident. I feel that this is an important initiative from the perspective of improving automated driving and other qualities. However, the JAMA, which is a manufacturer, would like to organize the discussion by considering the following points. First of all, I would like to change the wording to cover the handling of data at the time of an accident and data other than the accident to prevent a recurrence. With regard to the data of so-called close calls, I would like to have a discussion on the purpose of use of each data, the scope of data to be acquired, its specifications, and technical means for sharing and storing it, including whose data will be acquired by whom and with whom it will be shared, and to work on it after clarifying it in advance. I will talk about the details of the data later, but I would like you to consider this point. Next, we believe that it is necessary to make it a requirement to acquire excessive information that is not technically predictable, and to thoroughly implement rules that do not require disclosure beyond the protection of the rights of data owners. The issue of Haneda Airport has become a topic of conversation. Considering the current laws in Japan, if the provision of information to investigative authorities is directly adopted as evidence and leads to prosecution, it does not necessarily protect rights, and accident analysis can be an obstacle from the perspective of preventing recurrence. This is connected to what Mr. Suda also mentioned. If the manufacturer is required to collect unlimited data and disclose data without restrictions, the cost increase and unacceptable risk increase more than necessary, which may limit the spread and expansion. I would also like you to handle these points carefully.

Next, regarding data cooperation, as Dr. Nishinari said earlier, there are various forms of data related to automated driving. Even if we only discuss the matter of simply using and sharing it, I believe that it does not actually approach social implementation. When it comes to automated driving, the types and specifications of data necessary for cause analysis are not the same in each step of recognition, judgment, and action. I believe that it is essential to clarify this through continuous consultation with design and the manufacturing side. For example, considering the certainty of recognition, it is essential to obtain data in the form of a standard or in combination with data to be compared. Since it is impossible to confirm the correctness of a single data, a mechanism that ensures redundancy is required. When it comes to the certainty of judgment, not only external cognitive information such as sensors but also internal calculation data, i.e., records of how programs operate, are required, and in some cases, disclosure of confidential information, which is a source of competition, is required, so I believe that careful discussion is necessary. In addition, while there is extremely precise data such as collision waveforms in vehicle behavior, all of the data is not necessarily necessary for accident investigation, so I believe that it is necessary to clarify the scope of the record, the life of the data, and the discussion of when to discard the data.

My third point is not limited to the initiatives such as the Digital Zenso and the automated driving Support Road introduced by Mr. Ouchi, but in relation to the various measures being considered by the mobility Working Group and the points at issue this time, the JAMA consistently believes that automated driving needs a safety system that integrates the three issues of people, cars, and the traffic environment. However, if we do not clarify the functions to be performed by each of them and the responsibility demarcation associated with them, and share those standards, including the selection of means, I assume that it will not be linked to public activities. The means to achieve this is not limited to ITS, but it is known that automated driving safety can be improved to a considerable level even with analog measures such as installing right turn signals at intersections and changing the timing of pedestrian and vehicle separation, so I would like you to consider such a matter. In that case, in the overall image of the digital lifeline, we would like to clarify the responsibilities and functions to be performed by the traffic environment and discuss the definition of data that matches it.

Finally, I would like to give you an overview of the opinions. As you mentioned at the beginning, I understand that there are so many points of discussion and opinions from so many experts. Therefore, I think it is very difficult to conclude each point of discussion in each subworking session. The JAMA proposes that we discuss each point in parallel and consider whether we can spend enough time to consider them. There are four major points of discussion: civil, criminal, administrative, and accident investigation. On the other hand, I am the only one from the industrial world who can comment as a manufacturer. So, I would like you to consider how to proceed with the discussion in order to have as many discussions as possible. That is all from the JAMA.

YOKOTA Member: Insurance Co., Ltd. is actually engaged in the payment of insurance claims for automobile insurance or compulsory automobile liability insurance. In such a situation, it is extremely important to understand the causes of accidents. We believe that it is a topic of high interest to the General Insurance Association of Japan to collect and observe the mechanism for investigating the causes of accidents when Level 4 automated driving vehicles accidents occur, the travel data of vehicles necessary for that purpose, and the data on the surrounding environment.

First, in the case of manual driving by a driver, in what form do you conduct a damage investigation? We look at the damage to the vehicle as a confirmation of the accident site. In addition to that, hearing from drivers is very important, and we are taking measures against accidents based on that. Specifically, we ask the driver how he was driving, how the car was moving, how he collided with the other party, and so on, and determine whether or not it is covered by the insurance. In addition, in the case of a car-to-car accident, we investigate the percentage of negligence with the other party, and in the case of a single vehicle accident, we investigate whether it is the driver's operation error or the cause of the car, and we actually pay the insurance money. This is the current practice of manual driving, but when we think about what will happen if the system becomes the main driving system in the future, one possibility is that even if there is a driver, if an accident occurs, he or she is not very aware of the surrounding situation. Or, in the case of unmanned operation, it will be difficult to investigate the accident by interviewing the driver. Therefore, we believe that it is necessary to take appropriate measures after investigating the cause of the accident through the use of various data. Of course, when acquiring data, it is necessary to determine the purpose of use of the data and the scope of data required in the future. Even if the number of accidents is small at first, it is conceivable that the number of accidents will increase in automated driving society. Therefore, it is necessary to speed up the investigation of the causes of accidents by opening up data and standardizing data standards. I would like to discuss with you how to handle data in the future.

Takahashi Member: I am a lawyer, but I originally graduated from a science university, so I am a person who has a strong idea of science. I have a lot of experience in such cases, and I mainly deal with medical care error cases and cases that require engineering examination even in traffic accidents. From that perspective, I have a basic stance that the development of science and technology is indeed necessary, and that it would be disadvantageous for the people to hinder it. When I looked at it from such a perspective, the important data example of automated driving vehicles on page 2 of Dr. Nishinari's material was very helpful. In particular, I think it is very important how cars and people move and how to predict five seconds later. The reason is that when humans drive, how they judge things is that they first recognize information from the outside world, compare that information with past information that has already been stored, recognize and judge what it is, and then predict how to move based on that judgment. In humans, such predictive calculations are basically controlled by the cerebellum. The same thing is needed in automated driving. In my experience, the most traffic accidents are accidents on a pedestrian crossing without traffic control on a single road, accidents on a pedestrian crossing when turning right or left at an intersection with traffic control, and accidents at an intersection with poor visibility without traffic control. If automated driving vehicles observes the Road Traffic Law 100 percent, there will be no accidents. The Road Traffic Act is so precise. However, if there is no automated driving, it is against the Road Traffic Law. There are two main causes of accidents. The first is speeding. The second is a violation of the obligation to keep a close watch on the left and right in front of the vehicle. Where this happens is at crosswalks and intersections. Unfortunately, the number of vehicles that slow down or stop before a crosswalk is far less in Japan than in Europe. If there is a crosswalk in front of you on a straight road and there is no traffic light, you have to move at a speed that you can stop before the stop line. The reason is that even if a pedestrian who is going to cross suddenly starts to cross from behind something, it is not necessary to brake suddenly right before the pedestrian crossing, and it is possible to protect the pedestrian completely by stopping naturally. In other words, the traffic law is made on the premise that pedestrians who cross a crosswalk can trust that a car will approach them at a speed at which they can stop without sudden braking. Therefore, even in the case of a automated driving vehicles, unless it is obvious that there are no people at the crosswalk, it is necessary to drive at a speed that stops short of the stop line. By the way, in the case of automated driving, I think that it is impossible to achieve the judgment that it is clear that there are no people unless we further develop our technology. As I indicated last time, if there is a 1m hedge on the sidewalk on the left side of the crosswalk, an 80 cm elementary school student will be hidden there and will not be seen. If you want to find this, you have to use GPS to judge it in three dimensions. Recently, there was a case in which a child was hidden in a hedge, and when a car turned right at a green light, the child was hit on a crosswalk provided at the exit of an intersection. Such cases are not rare. At this point, I think it is quite doubtful that automated driving vehicles's technology has been taken into account to that extent. As Mr. Nishinari says, if you focus on safety, efficiency will decrease. As a typical example, if you obey the Road Traffic Law completely, there will be a big traffic jam on the Metropolitan Expressway. The speed limit of Metropolitan Expressway is 60 km / h. In addition, the same is true for ordinary highways. If it rains, the speed limit will be 50 km. I think that the Kobotoke Tunnel will be congested for about 100 km at 50 km. If the current road traffic law is completely observed like this, the efficiency may decrease. We have to solve this problem, too. On the other hand, if automated driving does not fully comply with the Road Traffic Law, the state will give its seal of approval to vehicles that violate the law driving around the world. When an accident occurs, I think it is impossible to convince the people that immunity from criminal prosecution will be granted. This is my first question.

Next, I would like to point out about Dr. Suda's materials. It is important to maintain records. In fact, current drive recorders are very incomplete. There have been three cases so far, but only the video of the collision has disappeared. Only the video of the most important collision has disappeared for several minutes before and after. It is not that the driver intentionally erased the video, but in fact, the police officers immediately seized it, but it is not technically clear why such an incomplete thing occurs. If it becomes a automated driving as it is, it will be difficult to determine the cause of the incident.

Another is that Mr. Suda's case of the Japan Marine Accident Tribunal has been partially integrated, and I think it is very good that an independent committee has been established in the Japan Transport Safety Board to some extent, but the authority is too weak. Several years ago, there was an accident in which 13 students fell from a ski bus in Karuizawa. I am working as a defense team for the victims, and the victims are also participating in criminal trials. Just recently, the dispatcher and the president of the company were sentenced to prison. At that time, both the Commercial Vehicle Accident Investigation Committee of the Japan Traffic Accident Analysis Center and the Nagano Prefectural Police conducted similar investigations. I'm sorry, but the investigation by the business automobile accident investigation committee at that time was insufficient. In response, the police are thorough, and actually stop traffic at the scene and make a demonstration experiments. The experiment clarified what the cause of the accident was. I was unfamiliar with the finger control transmission. For example, we shift from 5th to 4th, and from 4th to 3rd quickly, but with this finger-controlled transmission, when you shift from 5th to 4th, if you don't wait for 1 or 2 seconds before shifting to 3rd, the actual gear will be neutral even though the lever is physically in 3rd. The driver was unfamiliar with this. It seems that the Commercial Automobile Accident Investigation Commission basically conducts investigations based on hearings from the parties concerned in many cases, but unless the Commercial Automobile Accident Investigation Commission is granted the same degree of compulsory investigation authority as the police, I don't think the real cause of the accident can be found. I would like you to strengthen your authority.

Another thing is that I completely agree with Dr. Hatano's view that it is necessary to discuss each point in parallel and take sufficient time to consider it. There are various issues here, and we may discuss them next time, but we need to discuss them from the perspective of the protection of victims with regard to impunity. In addition, we must of course consider the protection of the rights of data owners, but at the same time, we must also consider the protection of the rights of crime victims. We must also consider how to harmonize them, and we cannot be biased toward one side only. I think we need to discuss them thoroughly and over sufficient time, and I would like you to stop making conclusions in these six discussions.

Mr. Suda: You are right that although you say it is not a place to pursue responsibility, there is a concern that it will be used to pursue responsibility. As I said last time, I think it would be good to have some kind of system like immunity from liability from a long-term perspective. Karuizawa was a matter for another organization, the Commercial Vehicle Accident Investigation Commission, not the Japan Transport Safety Board.

Takahashi Member: That's right. I'm sorry. I will correct that.

Mr. Suda: You are right that although you say it is not a place to pursue responsibility, there is a concern that it will be used to pursue responsibility. As I said last time, I think it would be good to have some kind of system like immunity from liability from a long-term perspective. Transport Safety Board covers railways, aviation, and ships, and that the Commercial Vehicle Accident Investigation Report Board operates under a different mechanism for automobiles.

Takahashi Member: . I understand.

Mr. Nakahara: Thank you very much for your very easy-to-understand presentation to the three members. I understand that there are various purposes for information sharing, and that if the information itself is not retained, it will be very difficult to investigate the cause of the incident. I would like to ask questions to each of the three members, but since I am a specialist in civil law and a complete outsider, I would like you to point out any errors in my recognition. When I looked at the provisions of Mr. Nishinari's German legal system, I think that sharing information leads to multiple purposes. Regarding the provisions (4) to (7), the first is confirmation of operational monitoring and compliance with obligations by public authorities, the second is safety research by research institutions, and the third is pursuing legal responsibility by victims in (7). I think Article 7, paragraph (1) cited there is a provision on pursuing civil responsibility against holders, and I understand that it is a system that aims to contribute to all of these by acquiring information for pursuing legal responsibility by victims. Compared with the presentation by Mr. Ouchi of DADC, I would like to ask how this system will be useful for automated driving vehicles to prevent accidents by other development manufacturers, or whether it will be realized by a completely separate mechanism.

Regarding the presentation by Mr. Ouchi of DADC, it is said that the collected data will not be used for the purpose of pursuing legal responsibility, or that the identification of the cause of the incident is out of scope. Rather than identifying the cause of the incident, I think it is an image of collecting various data to prevent the incident. At that time, we will have to collect a large number of types of data, and as stated on page 2 of the slide, it will be very important to make information collection and data provision mandatory or to provide it voluntarily. I would like to ask you in detail how you are strategically thinking about collecting information. In addition, on page 7, it is stated that you would like public authorities, the accident investigation agency, and others to use it in the future, but I feel that the purpose will be broader than the mechanism you are thinking about now, so I would like to ask what kind of specific things you are thinking about.

Regarding your presentation, Mr. Suda emphasized that the investigation of the accident was not aimed at pursuing legal responsibility. However, what I think from the fact that we could not obtain the consent of the guide this time is that if the report of the accident investigation is made public, the issue of responsibility will naturally come up. I would like to ask what Mr. Suda's thoughts are on the harmonization of the accident investigation and the pursuit of responsibility.

Mr. OUCHI: In terms of how to collect the first items, I believe that they will be collected voluntarily in the initial stage at the development stage. Like aircraft, I believe that depending on the magnitude of the accident, submission will be mandatory or voluntary.

Regarding the second point that the purpose will be expanded to the accident investigation agency, specifically, as a plan for the case where there is no one to testify because there is no driver, it is possible to create an environment model or something that can reproduce an accident in a virtual space from data and share it.

Ochiai Member: Based on Professor Inadani Each teacher made a presentation, which was very helpful, including surveys, systems, and operations in autonomous driving. The first point is what I felt in the presentation by the Japan Automobile Manufacturers Association. For example, in Mr. Nishinari's statement, I thought it was good in the basic direction to make it possible for cooperative areas to continuously take safety measures at multiple companies, not just one company. On the other hand, what I thought was important in my presentation was that it is necessary to maintain the scope of data use, acquisition, generation, and use in an appropriate design, which is called data governance. I thought it was important to clarify the scope of the agreement to a certain extent in advance. As stated in the DADC announcement, I thought it would be difficult to cooperate if the acquisition stage is voluntary and the submission is gradually requested. From that perspective, whether or not to make it the same as the German version that Dr. Nishinari explained, I think it is important to take the design of collecting data in some form from the first stage and making it available to each company in the industry to improve safety. I thought about the significance of data-sharing for a while, but of course, it will not be possible to resolve a case unless the cause of a final individual case itself is investigated. However, the information collected there is often limited to investigative authorities and the Japan Transport Safety Board, so it may not necessarily lead to information linkages within the industry. In addition, I believe that such a strong authority cannot generally be invoked for the sake of information linkages to the extent that it is not directly related to the incident. Therefore, I thought that two systems of data collection were necessary to investigate the cause of the incident from various information, beyond the limit of accident investigation to clarify individual cases.

Second, regarding the individual accident investigation that Dr. Takahashi and Dr. Suda were discussing, the technology of autonomous driving is also developing, and as Dr. Nishinari said, the contents of the presentation can change every week. Therefore, I think it would be good if specialized organizations and experts who are capable to a certain extent from such a perspective were involved as much as possible. On the other hand, regarding the authority of the accident investigation, although the discussion of data sharing and individual accident investigations are different, I think that the authority for data sharing within the industry must be voluntary and within a certain range. Voluntary means, for example, within the range specified by law, but in the case of individual accident investigations, there are no particular restrictions. Basically, it is a principle that the investigation agency can collect the information, and I feel that I have strong authority. I feel that it is necessary to discuss these separately. For example, as I was involved in another industry in the past, the Accident validation Conference under the Telecommunications Business Act has the former nature of information sharing, but the one that Dr. Suda talked about the operation seems to be the resolution of individual cases or intermediate in some cases, but I think that committees with each nature have already been used in combination with other systems, so I thought it would be appropriate to discuss them separately.

As for the third point, I think that you are absolutely right about how to think about the situation of the premise that Dr. Hatano talked about. I have been premised since the first time that it is impossible to organize various things without thinking about what is actually the responsibility demarcation point. For example, the premise regarding roads is being discussed in the Digital Zenso, etc., but I think it is socially acceptable to keep some sections of the Tomei Expressway off-limits, and in that case, the responsibility of automobiles is the same as that of streetcars running on the track unless you leave the lane. I thought that adding some restrictions such as that may lead to ensuring the safety of autonomous driving and improving social predictability at this point. As for what I would like to understand more deeply, I would like to ask Dr. Suda and Dr. Nishinari where the difference between general automobile accidents and autonomous driving traffic accidents is essentially. In addition, I believe that the accident investigation will be conducted from a scientific point of view, but I would like to ask how autonomous driving is positioned when compared with a scene in which more advanced systems are developed, including not only automobiles but also aviation.

Secretariat: I will read the written opinion of Mr. Inadani, who was absent today, on behalf of him. I have two comments. The first is about data connections and others. I have received an opinion that it is important for the government to establish standard specifications for safety-related data and to technically develop a mechanism that can be shared between the parties concerned, including information on close calls. I have received an opinion that it is important because the issue raised last time, whether to understand safety as a probabilistic thing or as something that behaves in the same way as humans, is related to the direction of system design and the ideal form of incentive design.

Secondly, we received opinions on the incentive design for investigation. In order to respond to the risks of advanced science and technology across the entire social system, it is essential to develop an investigation system to learn from the accident. It is pointed out that if the incentive design for companies related to the accident to be related to the investigation is insufficient, there is a possibility that the information necessary for improving the safety of the automated driving vehicles as a whole social system may not be obtained due to the limitations of the institutional capacity and human and physical capital of the investigation agency. Therefore, we received opinions that sufficient consideration should be made on whether sufficient incentives can be provided under the current law, taking into account that foreign companies may also be subject to accident investigation.

Takahashi Member: 's written opinion. I think that's true about the incentive design for sincerely providing data without hiding it. However, in order to move people or to make them tell the truth, I think both candies and whips are necessary. Candies are also necessary. However, I think incentives alone are not enough. Candies are also necessary. In such a case, I think that utilizing plea bargaining is one option. However, as I will say later, considering the feelings of damage, it is necessary to carefully consider it. If there was some kind of defect when the accident actually occurred, I think that there was some fault in the person in charge of the technical development. In such a case, instead of suspending the prosecution of the person in charge of the technical development through plea bargaining, if the person in charge of the technical development has a boss who has dominant authority, it is theoretically possible to impose punishment on both the boss and the organization. However, the current Japanese organization punishment is extremely lenient, and the fine is about 1 million yen to 2 million yen. Instead, it may be possible to make them provide data by imposing an enormous fine of a certain percentage of sales or profits, as in other countries.

Second, there is an expression of an investigation system to learn from accidents described in Professor Inadani's written opinion. It is natural to learn from actual accidents, but when thinking about how to prevent accidents from occurring, as I said earlier, 100% of the road traffic law will be observed. However, the road traffic law is written only in an abstract manner, and in many cases, the judicial precedents of criminal cases in the past are used as the standard. When there is no traffic control at an intersection with poor visibility, the standard for determining whether or not one road is a priority road with respect to the other intersecting road is a judicial precedent that if it is not more than 1.5 times the other intersecting road, it does not fall under the category of a priority road, and if it is more than 2 times the other intersecting road, it falls under the category of a priority road. However, the judicial precedents are very divided between 1.5 times and 2 times. Legal driving cannot be realized unless such criminal judicial precedents are collected using AI, etc. From the perspective of traffic accident victims, it is true that they would like to be punished, but what they want to know the most is the cause of the accident. However, this does not mean that they are considering the possibility of granting criminal immunity in exchange for clarifying the cause of the accident, so I would like you to keep this in mind. In addition, an enormous amount of organization punishment and data collection are necessary.

Member Sato: There can be various discussions about the range of data to be collected, but I think that collecting a wide range of data will contribute to the investigation of the cause. On the other hand, as stated in the opinion of the Japan Automobile Manufacturers Association, if too much information is collected, it will be a burden on industry. In the case of China that you introduced, for example, it was indicated that it is necessary to store data 90 seconds prior to the accident. However, I believe that it is also necessary to consider whether such data is really necessary and whether excessive data collection is required when considering the scope of data. In the same way, I felt that it would be quite difficult to obligate the storage and disclosure of data that is filled with the know-how of each manufacturer, not cooperative areas. Currently, based on international discussions, a standard for data storage devices for automated driving called DSSAD has been established, but if we go beyond that and ask only Japanese companies to obtain more detailed data, it will be an excessive regulation, and I think there is an argument that it could become a non-tariff barrier. Furthermore, as the level of automated driving rises, if there is a discussion in the UN to collect more detailed data than the current DSSAD, I think it is necessary to keep pace with it. On the other hand, it is possible to request the collection of information on incidents and the like as cooperative areas information, which will lead to the improvement of safety, so I think it is better to actively request it. We believe that the scope of data collection should be determined in consideration of the balance described above. From the perspective of accident investigation, it is very important that the investigation is conducted by a organization with sufficient authority, such as the Japan Transport Safety Board, as you mentioned, and that the results of the investigation are made public to further contribute to safety. On the other hand, the current setup of the Japan Transport Safety Board is that legal testimony, etc. will be forced, a report will be made, it will be made public, and the report itself will be used in trials, etc. I believe that it is necessary to further discuss whether it is appropriate to apply this to automated driving, while there are discussions on criminal immunity.

Member: Dr. Suda pointed out that unlike the Japan Transport Safety Board, which is based on law, the Investigation Committee on Accidents in automated driving and the Investigation Committee on Accidents in Commercial Vehicles as mentioned earlier do not exist based on law in the form of outsourcing to ITARDA, and I heard that the basis for what can be investigated is not clear. I believe that whether the information obtained from the investigation can be used to pursue criminal responsibility is a bigger issue, but I feel that there was no objection to the point that the authority needs to be strengthened.

There may be a problem of how to set this up and whether it will be incorporated into the Japan Transport Safety Board, but I don't think there are so many restrictions. If there are any difficult problems in the Ministry of Land, Infrastructure, Transport and Tourism, I would like to ask you. For example, the accident investigation targets of the Japan Transport Safety Board are listed, but if all traffic accidents are included in it, it may be too large. After all, the number of cases may be very large compared to railway accidents, aircraft accidents, and marine accidents, so I think the method of cutting the investigation targets will be discussed in the future.

It was very interesting to hear that Germany was going beyond the investigation of individual accidents to collect data more widely and make use of it. However, as pointed out by the Japan Automobile Manufacturers Association, the scope is too wide and it is not clear how it will lead to the pursuit of criminal responsibility. I thought it was reasonable to point out that as a business operator, we would have to be quite cautious. If such a law had been enacted in Germany, it is good that there is 1g of the German Road Traffic Law, which has come to a conclusion. However, I felt that it is necessary to thoroughly investigate the parts that were discussed in the process and from what perspective German automobile manufacturers agreed to create this law.

In addition, Dr. Nishinari pointed out that if German manufacturers are good enough in Germany, German-made cars may come to Japanin the future, and American and Chinese cars may also come to Japanin the future. Considering that foreign manufacturers will come to Japan in the future, Dr. Nishinari pointed out the need for a survey mechanism and data standardization. It may be good to do things within Japanin the present situation, but I feel that if there is no international harmonization, there is nothing we can do in the end. Therefore, I think we cannot do nothing, but I would like to hear information on how we should move forward. I am aware that automated driving is probably being discussed at the UN, and I would like to hear information on how this is going on, even if it is not today.

Finally, regarding the future course of action that you pointed out as the general remarks of the JAMA, I am also deeply concerned about how a conclusion will be reached in the next few months. I think it is very good that the Secretariat has come up with a plan for the direction and is avoiding considering it in the form of a slap, so that the direction will not be determined from this stage. On the other hand, I feel that I do not know if a conclusion will really be reached in another five months. Just before the meeting, the theme of "Next, we will do this" is presented, but if you show us a roadmap for how we plan to discuss in the future, I feel that we will be able to discuss in a little more depth than in the last few days. I think you are making great efforts, but I would like you to consider that point.

Secretariat: Transport Safety Board is the organization that has the legal authority to investigate accidents that occur in aviation, railways, and shipping based on law. In addition, the Commercial Automobile Accident Investigation Commission and the automated driving Accident Investigation Commission are separately established in ITARDA. As a background, the Transport Safety Board responds to accidents that occur in aviation, railways, etc., but while there are no experts on automobile accidents, ITARDA has such knowledge and experience, so I am aware that we have been asked to do so. I cannot answer here whether it is better for the Transport Safety Board to do so, but I am aware that there is a discussion on whether the authority of the Accident Investigation Commission is necessary and sufficient in its current form. In addition, I am aware that when there is a discussion that it is not good to investigate accidents in automated driving vehicles where there is a legal background, it will be a discussion on who should have the legal authority. It may be the Transport Safety Board and other organizations that have been discussed so far, but I would be very grateful if you could discuss them in that order.

Secretariat: We have received requests about why we cannot ride in an unmanned automated driving car, and we have been trying to hold careful discussions as much as possible. We hope to hold discussions on how to turn feedback loops in the macro and micro sectors using data. After this, we would like to hold a meeting once a month and have in-depth discussions on criminal, civil, and administrative issues. During that time, we may be able to make an interim report, but we would like to hold discussions with experts separately and proceed with administrative work in parallel. There were opinions from the Japan Automobile Manufacturers Association and Mr. Takahashi that we would like to have in-depth discussions more carefully, so we would like to have discussions in parallel, and finally, we would like to have the Secretariat make a proposal based on the discussions so far, and discuss it at the end. I would like to thank you for your cooperation in this urgent situation, including the adjustment of the schedule each time, but I would like to do it more thoroughly and carefully, so I would like to ask for your cooperation. Then, Mr. Ochiai has raised his hand, so I would like to ask for your cooperation. implementation

Ochiai Member: Based on Professor Inadani 's paper, I think that there are cases in which relationships with overseas business operators are not institutionalized, and if you make a voluntary request, it is difficult to get them to respond. In particular, IT companies can be part of the entities that have information, and if there is a possibility that information will be collected from overseas IT companies, etc., unless we consider the fact that it is specified in the system to a certain extent, I thought that there would be a concern that only Japanese companies will submit information and not foreign companies. However, I think that it is extremely important to consider what form can ensure effectiveness. I don't think it is healthy for only Japanese companies to follow.

In addition, as for another point, regarding the incentive for the investigation of the incident, I thought that Dr. Takahashi's opinion is actually close to Dr. Inadani's opinion. For example, there is a possibility that we will decide in what form we will cooperate in relation to the investigation of the incident or something like that in the future. There was talk of a fine, but I think there was also an issue of corporate sanctions. So, I thought that a certain reduction and exemption in the case of appropriate cooperation with such sanctions systems, and the plea agreement Dr. Takahashi talked about and Dr. Inadani's opinion on DPA were close in a broad sense.

Takahashi Member: As we discussed earlier, it is important to establish a separate investigation committee on accidents related to automated driving, although it was a separate committee in the case of the Japan Transport Safety Board, which is mainly engaged in aircraft accidents and maritime accidents, and in the case of Karuizawa Ski Bus. In fact, it is because considerable scientific and technical knowledge is required. I do not intend to say anything to the NPA, but I wonder whether the police and legal institutions are capable of implementing it. Since considerable specialized knowledge is required, I think it is necessary to create a new investigation committee on accidents related to automated driving, with scientific and technical experts, the NPA, the Ministry of Justice, and crime victims as members. In fact, the medical care Incident is moving in that direction. The medical care Error Case has developed considerably over the past 20 years, and there have been many precedents in the Supreme Court, so hospitals feel a sense of crisis. In most large hospitals, investigation committees have been established. Their results are also used to some extent in civil and criminal cases. There may be a debate over how much data from the Accident Investigation Committee should be used in criminal or civil cases, but there is no direction to not use it at all, and I think it is necessary to use it to some extent.

Next, regarding the talk about incentives and organization penalties that I mentioned earlier, I think it is one method if we consider the development of science and technology. Regarding organization penalties, if we pay tens of billions of yen or hundreds of billions of yen, instead of 1 million yen or 2 million yen, as in the U.S., it will be a fairly good sanction and an opportunity to provide data. On the other hand, in the case of Japan, punitive compensation in civil cases is not allowed in court cases. In fact, even if plaintiffs claim compensation for punitive penalties, they are always dismissed. Therefore, at least in civil cases, I think punitive penalties are impossible unless laws are developed. On the other hand, if we impose hundreds of billions of yen in criminal organization penalties, it is also very difficult for victims to be convinced. It is difficult to exempt persons in charge of development in technology from criminal punishment even when they were grossly negligent or intentionally negligent. To put it the other way around, criminal immunity will be granted when there was normal negligence, but I think this is also difficult considering the feelings of victims. I think we need to make adjustments to reach an agreement on that.

Mr. Kozuka: Thank you very much. We would like to close the meeting here. After that, if you have any additional opinions, we would like to receive them from the Secretariat by the end of this week. In addition, the materials for today's meeting will be published on the Digital Agency website at a later date. As for the minutes, we would like to publish them on the Digital Agency website after confirming the contents with the members, as we did last time, so we would like to ask for your cooperation. The next sub-working group meeting is scheduled for February 27. The agenda is currently under consideration, and I would like to ask for your cooperation even though it may be a burden on you. Then, we will close the second sub-working group meeting on the ideal social rules for automated driving vehicles in age of AI. Thank you for taking the time to participate. . Today, I received a variety of opinions. Overall, I heard that there were no objections to the direction of sharing safety-related data throughout society and eliminating accidents, and that it was cooperative areas, not competitive areas. However, there is a big problem of how to implementation there. On the one hand, there are technical issues, and today, I received a presentation on what kind of information, in what form, how to format it, and what kind of platform to put it on. The teachers' talk was exactly related to that, and the secretariat mentioned it in the middle, but I felt that it was necessary to have a careful dialogue with the people concerned and work to narrow down what could be narrowed down. On the other hand, there is also the issue of the authority of organization and how the actions of the parties concerned are controlled in terms of incentives, including punishment, and whether it is possible to move only in automated driving. Regarding Mr. Takahashi's second remark, there is a discussion that civil judgments should be shared in an open manner, and I think this will be realized, but criminal judgments will be another big discussion further down the road, and I think that this will be a problem to be raised in this subworking, but it is also important to create a safer world in automated driving through such a big problem, so I feel that it is necessary to raise possible problems. I would like the Secretariat to summarize various issues and opinions.

Finally, as some of you said, I don't think it will be a problem limited to Japanese manufacturers or only Japanese manufacturers. Japanese manufacturers will also go overseas, and I think automated driving vehicles of foreign manufacturers will also enter the Japanese market, so I felt that I would like to not forget such a perspective. It's about time, so finally, I would like to ask Deputy Minister for Foreign Affairs of Digital Agency Mr. Yoshihisa Hasui to summarize today's situation. Mr. Hasui, could you please?

Mr. Hasui: Thank you, Mr. . Thank you very much for coming today even though you are busy, for giving me a last-minute guidance on the meeting, and for giving me various materials. Today, I received various opinions, including teacher's presentations, on the Issue and direction of the current systems for accident investigation, information sharing, and data connections. With regard to the investigation of the incident, Dr. Suda and others pointed out that a neutral organization with legal investigation authority is necessary for prompt and effective investigation of the cause. As Dr. Takahashi said, we should strengthen our authority, and we received very important points such as that we should create an investigation agency for automated driving. Based on these points, as I mentioned earlier from the Secretariat, we would like to cooperate with the relevant ministries and agencies and further deepen the discussion. At the same time, we received various opinions from Mr. Ouchi and Dr. Nishinari of DADC and others about validation to strengthen data connections for macro incidents, including minor accidents and close calls. I think this is an indication of the importance of how to identify the necessary information and data items according to the purpose of use of the information and data and how to share it with the people who need it. We will consider how to report and share based on this. Furthermore, from the perspective of foreign countries, it is common sense and common practice that data, in particular, transcends national borders. Based on this point, I believe that how to proceed with the meeting is also important. As we heard from the Secretariat earlier, we would like to continue our careful and in-depth discussions with the continued support of all parties concerned, including you and relevant ministries and agencies. I would like to summarize the themes including the points discussed this time in June based on the opinions of the teachers, so I would like to ask for your continued support in order to obtain various knowledge. Thank you for your time today.