Accident analysis system, accident analysis program, accident analysis method, and fault ratio estimation system

Abstract

To enable the specific identification of the causes of traffic accidents involving driverless, self-driving vehicles. [Solution] The accident analysis system (100) includes a first acquisition unit (131) that acquires time information indicating the time, location information indicating the position of the autonomous vehicle (5), and function information relating to multiple autonomous driving functions of the autonomous vehicle; a second acquisition unit (132) that acquires detection results from infrastructure sensors (3) that are placed along the autonomous vehicle's driving route and detect traffic conditions; a data generation unit (133) that generates reconstruction data of traffic conditions prior to the time of the accident based on time information, location information, function information, and detection results corresponding to the time of the traffic accident involving the autonomous vehicle; a function identification unit (134) that analyzes the reconstruction data to identify the problematic autonomous driving function among the multiple autonomous driving functions; and an output unit (11) that outputs information relating to the identified autonomous driving function.

Description

Technical Field

[0001] The present disclosure relates to an accident analysis system, an accident analysis program, an accident analysis method, and a fault ratio estimation system.

Background Art

[0002] Patent Document 1 describes an information processing device that derives the fault ratio of an accident using the driver's biometric information.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Means for Solving the Problems

[0004] In order to solve the above problems, an accident analysis system according to one aspect of the present disclosure includes a first acquisition unit that acquires time information indicating time, position information indicating the position of an autonomous vehicle, and function information regarding a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition unit that acquires a detection result by an infrastructure sensor disposed on a travel route of the autonomous vehicle and detecting a traffic-related situation; a data generation unit that generates reproduction data of a traffic situation before the occurrence of the traffic accident based on the time information, the position information, the function information, and the detection result corresponding to the time when the traffic accident in which the autonomous vehicle is involved occurs; a function identification unit that analyzes the reproduction data and identifies an autonomous driving function having a problem among the plurality of autonomous driving functions; and an output unit that outputs information regarding the identified autonomous driving function.

[0005] Furthermore, an accident analysis program according to another aspect of the present disclosure causes a computer to perform a first acquisition process that acquires time information indicating the time, location information indicating the position of the autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition process that acquires detection results prior to the time indicated by the time information from infrastructure sensors that are placed along the driving route of the autonomous vehicle and detect traffic conditions; a data generation process that generates reconstruction data of the traffic conditions prior to the time of the traffic accident based on the time information, location information, function information, and detection results corresponding to the time when a traffic accident involving the autonomous vehicle occurred; a function identification process that analyzes the reconstruction data to identify the autonomous driving function that had a problem among the plurality of autonomous driving functions; and an output process that outputs information relating to the identified autonomous driving function.

[0006] Furthermore, an accident analysis method relating to another aspect of the present disclosure includes: a first acquisition step in which a computer acquires time information indicating a time, location information indicating the location of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition step in which the computer acquires detection results prior to the time indicated by the time information from infrastructure sensors that are placed along the driving path of the autonomous vehicle and detect traffic conditions; a data generation step in which the computer generates reconstruction data of traffic conditions prior to the time of the traffic accident based on the time information, location information, function information, and detection results corresponding to the time when a traffic accident involving the autonomous vehicle occurred; a function identification step in which the computer analyzes the reconstruction data to identify the autonomous driving function that had a problem among the plurality of autonomous driving functions; and an output step in which the computer outputs information relating to the identified autonomous driving function.

[0007] Furthermore, a fault ratio estimation system according to another aspect of the present disclosure includes: a first acquisition unit that acquires time information indicating a time, location information indicating the position of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition unit that acquires detection results prior to the time indicated by the time information from infrastructure sensors arranged along the autonomous vehicle's travel route and detecting traffic conditions; an information generation unit that generates overview information showing an overview of traffic conditions from a predetermined time before the traffic accident occurred, based on the time information, location information, function information, and detection results corresponding to the time a traffic accident involving the autonomous vehicle occurred; a ratio estimation unit that inputs the overview information into a trained model constructed to output a numerical value when traffic information relating to traffic conditions before the traffic accident is input, and acquires the numerical value output by the trained model as the fault ratio of the autonomous vehicle; and an output unit that outputs the fault ratio. [Brief explanation of the drawing]

[0008] [Figure 1] This block diagram shows an example of a schematic configuration of an accident analysis system according to the embodiment of Disclosure 1. [Figure 2] This block diagram shows an example of the functional configuration of the accident analysis device included in the system. [Figure 3] This is a flowchart illustrating an example of the processing flow performed by the device. [Figure 4] This diagram illustrates some of the processes performed by the device. [Figure 5] This diagram illustrates some of the processes performed by the device. [Figure 6] This is a schematic diagram showing an example of the reproduction data generated by the device. [Figure 7] This table shows examples of the parties involved in each type of accident, as well as cases in which negligence may be present in those parties. [Figure 8] This is a block diagram showing an example of the schematic configuration of the fault ratio estimation system according to the embodiment of Disclosure 2. [Figure 9] This block diagram shows an example of the functional configuration of the fault ratio estimation device included in the system. [Figure 10] This is a flowchart illustrating an example of the processing flow performed by the device. [Figure 11] This table shows how to utilize the fault percentage and reason output by the device for each use case, as well as examples of output. [Modes for carrying out the invention]

[0009] <Disclosure 1> Disclosure 1 is explained below.

[0010] [Background, challenges, solutions, and principles of solutions] Prior to describing the embodiments of Disclosure 1, the background of Disclosure 1, the problems of the prior art, the means for solving the problems, and the principle of solving them by the means will be explained.

[0011] 〔background〕 As the number of self-driving vehicles increases in the future, there is a possibility that traffic accidents involving these vehicles will occur. If a self-driving vehicle is involved in a traffic accident, it will be necessary to determine the cause of the accident and, if so, whether there was negligence or the degree of negligence, just as it would be if a human-driven vehicle were involved in an accident.

[0012] 〔assignment〕 Conventionally, various technologies for estimating the cause of accidents have existed, such as those described in Patent Document 1 above. However, conventional technologies used the driver's biometric information. In other words, conventional technologies were only able to estimate the cause of accidents when a vehicle driven by a human driver was involved in a traffic accident. Therefore, conventional technologies had the problem of not being able to deal with traffic accidents involving autonomous vehicles that do not have a driver.

[0013] Furthermore, autonomous driving by self-driving vehicles is realized through collaboration among multiple parties (vehicle manufacturers, governments, operators, etc.). Therefore, if the self-driving vehicle is at fault, it becomes necessary to determine which party should bear responsibility for the fault. However, conventional technology cannot handle the determination of which party should bear responsibility for the fault.

[0014] [Solution means] In order to solve the above problems, an accident analysis system according to an aspect of the present disclosure includes a first acquisition unit that acquires time information indicating time, position information indicating the position of an autonomous vehicle, and function information regarding a plurality of autonomous driving functions of the autonomous vehicle, a second acquisition unit that acquires a detection result by an infrastructure sensor disposed on a driving route of the autonomous vehicle and detecting a traffic-related situation, and based on the time information, the position information, the function information, and the detection result corresponding to the time when a traffic accident in which the autonomous vehicle is involved occurs, a data generation unit that generates reproduction data of the traffic situation before the occurrence of the traffic accident, a function identification unit that analyzes the reproduction data and identifies a problematic autonomous driving function among the plurality of autonomous driving functions, and an output unit that outputs information regarding the identified autonomous driving function.

[0015] [Solution principle] The accident analysis system configured as described above generates highly accurate reproduction data in consideration of objective information such as the detection result of the infrastructure sensor. Then, the accident analysis system can identify the cause of the accident by analyzing the reproduction data. Also, by analyzing the reproduction data in more detail, when there is a problem somewhere in the autonomous vehicle, it is also possible to identify whether a traffic accident such as the reproduction data will occur. The accident analysis system according to Disclosure 1 can thus identify the cause of a traffic accident involving an autonomous vehicle without a driver.

[0016] [Embodiment of Disclosure 1] Next, an embodiment of Disclosure 1 will be described in detail with reference to the drawings.

[0017] [Schematic configuration of accident analysis system 100] As shown in Figure 1, the accident analysis system 100 comprises an accident analysis device 1 and a detection result database 2. The accident analysis system 100 is also connected via a network N to at least one infrastructure sensor 3, at least one traffic management device 4, at least one autonomous vehicle 5, a remote monitoring device 6, and other systems 7. Note that the detection result database 2 does not have to be included in the accident analysis system 100. The detection result database 2 may also be configured integrally with the accident analysis device 1. At least one of the at least one traffic management device 4, at least one infrastructure sensor 3, at least one autonomous vehicle 5, a remote monitoring device 6, and other systems 7 may be included in the accident analysis system 100. The accident analysis system 100 may also be incorporated into any of the autonomous vehicle 5, the remote monitoring device 6, and other systems 7.

[0018] (Infrastructure Sensor 3) The infrastructure sensor 3 is positioned along the driving path of the autonomous vehicle. The infrastructure sensor 3 detects traffic-related conditions. The "driving path" includes the road surface of the driving path, the area above the driving path, the sides of the driving path, and buildings facing the driving path. The "infrastructure sensor 3" includes LiDAR (Light Detection And Ranging), cameras (image sensors), radar, etc. LiDAR and radar detect the distance to each vehicle, pedestrian, obstacle, etc. within their detection range. Cameras photograph vehicles, pedestrians, obstacles, etc. Note that the infrastructure sensor 3 may be an on-board sensor installed in a vehicle other than the autonomous vehicle 5 that is driving along the driving path. In this embodiment, the infrastructure sensor 3 repeatedly detects traffic-related conditions at predetermined intervals. If there are multiple types of infrastructure sensors 3, each infrastructure sensor 3 repeatedly detects traffic-related conditions at different predetermined intervals. When the infrastructure sensor 3 detects traffic-related conditions, it transmits the detection result to the accident analysis device 1.

[0019] (Traffic control device 4) The traffic management device 4 is a device that manages the movement of various vehicles, including the autonomous vehicle 5. The traffic management device 4 includes traffic signals, railway barriers and warning devices, various electronic traffic information display devices, etc. The traffic management device 4 transmits a traffic management signal indicating the action it has taken to the accident analysis device 1 every predetermined time or every time it switches an operation.

[0020] (Autonomous vehicle 5) The autonomous vehicle 5 is equipped with various on-board sensors. The autonomous vehicle 5 automatically travels along a pre-set route while repeatedly performing detection, recognition, judgment, and control. "Detection" is when the various on-board sensors detect the conditions inside and outside the autonomous vehicle 5. "Recognition" is when the autonomous vehicle 5 recognizes the presence of surrounding objects, road surface conditions, etc., based on the detection results of the various on-board sensors. "Judgment" is when the autonomous vehicle 5 decides what action to take (acceleration / deceleration, turning left or right, etc.) based on the results of the recognition. "Control" is when the autonomous vehicle 5 controls the drive mechanism, steering mechanism, etc., equipped with the autonomous vehicle 5 based on the results of the judgment.

[0021] Furthermore, the autonomous vehicle 5 generates a vehicle log showing various actions it has performed at predetermined intervals and transmits it to the accident analysis device 1 along with time information. The "vehicle log" includes logs related to space (perception, sound), planning, and behavior. The logs related to space (perception, sound), planning, and behavior each include logs related to hardware, software, etc. The "time information" is information (timestamp) indicating the time when the various vehicle logs were generated.

[0022] Among the "logs related to space (cognition, sound)," "logs related to hardware" include logs related to in-vehicle sensors, GNSS (Global Navigation Satellite System) logs, and high-resolution map logs. "Logs related to sensors" include logs related to cameras, LiDAR logs, and radar logs. Among the "logs related to space (cognition, sound)," "logs related to software" include logs related to sensor fusion, self-localization logs, and cognitive AI logs.

[0023] "Logs related to planning" include "logs related to hardware," which include logs related to operation plans, etc. "Logs related to operation plans" include logs related to ECUs (Electronic Control Units), etc. "Logs related to planning" include "logs related to software," which include logs related to operation plans, etc. "Logs related to operation plans" include logs related to risk calculations, etc.

[0024] Among the "behavior logs," the "hardware logs" include logs related to following the vehicle ahead, logs related to lane following, etc. The "logs related to following the vehicle ahead" include logs related to the ECU, logs related to on-board sensors, etc. Among the "behavior logs," the "software logs" include logs related to lane following, logs related to following the vehicle ahead, etc. The "logs related to lane following" include logs related to following calculations, logs related to steering control, logs related to instruction value calculations, etc.

[0025] These vehicle logs each include information such as whether or not any of the autonomous driving functions of the autonomous vehicle 5 were performed correctly, or information obtained as a result of the performance of autonomous driving functions (functions to "recognize" objects, functions to "judge" actions, functions to "control" drive mechanisms, functions to "detect" various on-board sensors). In other words, the vehicle logs are also functional information related to the multiple autonomous driving functions of the autonomous vehicle 5. Furthermore, among these vehicle logs, the logs related to GNSS include location information. "Location information" is information indicating the position of the autonomous vehicle 5. That is, the autonomous vehicle 5 in this embodiment generates location information using GNSS.

[0026] (Remote monitoring device 6) The remote monitoring device 6 monitors the operation of the autonomous vehicle. The remote monitoring device 6 also remotely controls the autonomous vehicle 5 as needed.

[0027] The remote monitoring device 6 generates monitoring logs showing various operations it has performed each time a predetermined period of time has elapsed, or each time it controls the autonomous vehicle 5, and transmits these logs to the accident analysis device 1. The "monitoring logs" include logs related to communication, logs related to driving environment conditions (Operational Design Domain: ODD), logs related to alerts, logs related to countermeasures, logs related to maintenance, etc. These monitoring logs, like the vehicle logs, also contain functional information related to the multiple autonomous driving functions of the autonomous vehicle 5.

[0028] "Logs related to communication" include logs related to throughput, logs related to communication status, etc.

[0029] "Logs related to ODD" include logs related to road conditions, geographical conditions, environmental conditions, and operating conditions. "Logs related to road conditions" include logs related to road type and number of lanes.

[0030] "Alert-related logs" include logs regarding whether or not an alert was triggered, logs regarding the timing of the alert, etc.

[0031] "Maintenance logs" include logs related to hardware, logs related to software, etc. "Hardware logs" include logs related to the maintenance and replacement status of various parts (tires, etc.) and logs related to the maintenance and replacement status of various consumables (brake fluid, etc.). "Software logs" include logs related to version upgrades, etc.

[0032] (Other systems 7) Other systems 7 are systems used by organizations involved in traffic accidents. These organizations include, for example, the police and insurance companies. The police are involved in conducting on-site investigations of traffic accidents and creating accident records. Insurance companies are involved in calculating the percentage of fault of the parties involved in a traffic accident and paying insurance benefits.

[0033] (Detection Result Database 2) The detection result database 2 stores the detection results received by the accident analysis device 1 from the infrastructure sensor 3. In this embodiment, the detection result database 2 also stores traffic management information received by the accident analysis device 1 from the traffic management device 4.

[0034] [Specific configuration of accident analysis device 1] As shown in Figure 2, the accident analysis device 1 comprises a communication unit 11, a storage unit 12, and a calculation unit 13.

[0035] (Communications Section 11) The communication unit 11 communicates with the detection result database 2, the traffic management device 4, the infrastructure sensor 3, and the autonomous vehicle 5, respectively. The communication unit 11 according to this embodiment also communicates with the remote monitoring device 6 and the other systems 7, respectively. Furthermore, the communication unit 11 according to this embodiment is composed of a communication module. In addition, the communication unit 11 according to this embodiment also serves as the output unit and the second output unit in Disclosure 1.

[0036] (Storage unit 12) The storage unit 12 stores the accident analysis program 121. The accident analysis program 121 is a program that causes the computer to function as an accident analysis device 1. The storage unit 12 may also be configured to store various calculation results from the arithmetic unit 13. In this embodiment, the storage unit 12 is composed of semiconductor memory, a hard disk drive, or the like.

[0037] (Computation unit 13) The calculation unit 13 comprises a first acquisition unit 131, a second acquisition unit 132, a data generation unit 133, and a function identification unit 134. The calculation unit 13 according to this embodiment further comprises an accident occurrence determination unit 135, a table creation unit 136, a first information generation unit 137, a second information generation unit 138, a negligence determination unit 139, and an output processing unit 140. The calculation unit 13 according to this embodiment is composed of a processor and memory. In other words, the accident analysis device 1 is composed of a computer. Therefore, the functions of each control block 131 to 140 are realized by the calculation unit 13 executing various processes S1 to S10 in the flow shown in Figure 3 according to the accident analysis program 121 stored in the storage unit 12.

[0038] • First acquisition process S1, first acquisition unit 131 In the initial first acquisition process S1, the first acquisition unit 131 acquires time information, location information, and function information. In this embodiment, the first acquisition unit 131 acquires time information, location information, and function information from the autonomous vehicle 5. As described above, the autonomous vehicle 5 transmits time information, location information, and function information every predetermined time. Therefore, in this embodiment, the first acquisition unit 131 acquires these each time the autonomous vehicle 5 transmits time information, location information, and function information. The first acquisition unit 131 may acquire at least one of the various types of information directly from the generating device (for example, the autonomous vehicle for time information, or the remote monitoring device for monitoring logs), or it may indirectly acquire information acquired and held by another device.

[0039] Furthermore, the first acquisition unit 131 according to this embodiment acquires at least one of the vehicle log and the monitoring log. As described above, the autonomous vehicle 5 transmits the vehicle log every predetermined time. The remote monitoring device 6 transmits the monitoring log every predetermined time or every time it controls the autonomous vehicle 5. Therefore, the first acquisition unit 131 according to this embodiment acquires these logs every time the autonomous vehicle 5 transmits the vehicle log and every time the remote monitoring device 6 transmits the monitoring log. The first acquisition unit 131 may acquire at least one of the various logs directly from the generating device (the autonomous vehicle for time information, the remote monitoring device for monitoring logs), or it may indirectly acquire logs acquired and held by other devices.

[0040] ·Second acquisition processing S2, second acquisition unit 132 After the first acquisition process S1, the process moves to the second acquisition process S2. In the second acquisition process S2, the second acquisition unit 132 acquires the detection result from the infrastructure sensor 3. As described above, the infrastructure sensor 3 transmits the detection result every predetermined time. If there are multiple types of infrastructure sensors 3, each infrastructure sensor 3 transmits the result every predetermined time, which differs for each type. Therefore, the first acquisition unit 131 in this embodiment acquires the detection result from the infrastructure sensor 3 every time it transmits it.

[0041] Furthermore, the second acquisition unit 132 according to this embodiment acquires traffic management information output by the traffic management device 4 on the travel route. As described above, the traffic management device 4 transmits traffic management information every time a predetermined period of time has elapsed or every time it switches operation. If there are multiple types of traffic management devices 4, each traffic management device 4 transmits the information every time a different predetermined period of time has elapsed or every time it switches operation. Therefore, the second acquisition unit 132 according to this embodiment acquires the traffic management information every time the traffic management device 4 transmits it. The second acquisition unit 132 may acquire the detection result directly from the infrastructure sensor 3, or it may acquire it indirectly from another device that has acquired and is holding it. Also, the second acquisition unit 132 may execute the second acquisition process S2 before the first acquisition process S1, or it may execute it in parallel with the first acquisition process S1.

[0042] Accident occurrence determination process S5, accident occurrence determination unit 135 The accident occurrence determination process S5 is performed while the first acquisition process S1 and the second acquisition process are being repeated. In the accident occurrence determination process S5, the accident occurrence determination unit 135 determines whether or not a traffic accident involving an autonomous vehicle has occurred. A "traffic accident involving an autonomous vehicle" includes both traffic accidents in which the autonomous vehicle 5 collides with another vehicle (caused by the autonomous vehicle 5) and traffic accidents in which the autonomous vehicle 5 is hit by another vehicle (involved in the autonomous vehicle 5). The accident occurrence determination unit 135 determines, for example, that a traffic accident involving an autonomous vehicle has occurred if the most recently acquired vehicle log shows an abnormal value, or if it receives a notification from the autonomous vehicle 5 or the remote monitoring device 6. In this embodiment, the accident occurrence determination unit 135 determines whether or not a traffic accident involving an autonomous vehicle has occurred each predetermined time interval.

[0043] • Table creation process S6, Table creation unit 136 When the accident occurrence determination unit 135 determines that a traffic accident involving the autonomous vehicle has occurred, the process proceeds to the table creation process S6. In the table creation process S6, the table creation unit 136 creates a data table. Specifically, the table creation unit 136 first converts the detection results for a plurality of times obtained so far and the traffic management information into a synchronized state, including the detection results corresponding to the time t when the traffic accident involving the autonomous vehicle occurred and the traffic management information. The "information corresponding to time t" includes information generated at time t, information generated at the time immediately before time t, and information generated at the time immediately after time t. As described above, the second acquisition unit 132 according to the present embodiment also acquires traffic management information. Therefore, the table creation unit 136 according to the present embodiment also extracts the traffic management information corresponding to the time t when the traffic accident occurred.

[0044] For example, as shown in FIG. 4, it is assumed that the detection results for a plurality of times t1, t2,... and the traffic management information are stored in the detection result database 2. And it is assumed that a traffic accident occurs at a certain position (coordinate) at time t (t5 < t < t6 in FIG. 4). In this case, the table creation unit 136 extracts at least the detection results (the position of the traffic accident occurrence location by the infrastructure sensor 3) and the traffic management information (the state of the traffic management device) corresponding to the times before time t (the times for which the first and second overhead views are to be obtained: for example, the times around time t, or the time t5 immediately before time t, and the time t4 one cycle before that, and the time t6 immediately after time t, and the time t7 one cycle after that, etc.). Then, the table creation unit 136 resamples (for example, linear interpolation, etc.) the detection results and the traffic management information corresponding to the times around time t. Thereby, a data table is created in which the values of various types of information with slightly different generation times are converted into values generated exactly at the times around time t.

[0045] · First information generation process S7, first information generation unit 137 If the accident occurrence determination unit 135 determines that a traffic accident involving an autonomous vehicle has occurred, the process moves to the first information generation process S7. In the first information generation process S7, the first information generation unit 137 generates first overview information that shows an overview of the traffic situation from the time of the traffic accident to a predetermined time prior, based on time information, location information, and function information. In this embodiment, the first information generation unit 137 generates, for example, a plan view showing the situation at a certain moment before the accident as the first overview information. As described above, the function information is information about multiple autonomous driving functions of the autonomous vehicle 5, and the autonomous driving functions include functions that are "detected" by on-board sensors. For this reason, in addition to the first overview information based on autonomous driving functions (cognition, judgment, control), the first information generation unit 137 in this embodiment also generates first overview information based on the detection results of the on-board sensors. Furthermore, the first information generation unit 137 in this embodiment generates first overview information for multiple time periods. The first information generation unit 137 may execute the first information generation process S7 after the first acquisition process S1 and before the second acquisition process S2, or it may execute it in parallel with the second acquisition process S2.

[0046] • Second information generation process S8, second information generation unit 138 After the table creation process S6, the process moves to the second information generation process S8. The second information generation unit 138 generates second overview information that shows an overview of the traffic situation from the time of the traffic accident to a predetermined time prior, based on the time information, location information, and detection results (data table). The second information generation unit 138 generates the second overview information in the same manner as the first information generation unit 137 generates the first overview information. In addition, the second information generation unit 138 in this embodiment generates second overview information for multiple time periods. The second information generation unit 138 may execute the second information generation process S8 before the first information generation process S7, or it may execute it in parallel with the first information generation process S7.

[0047] • Negligence determination processing S9, negligence determination unit 139 After the second information generation process S8, the process moves to the negligence judgment process S9. In the negligence judgment process S9, the negligence judgment unit 139 compares the first overview information and the second overview information to determine whether or not the autonomous vehicle was at fault. The "autonomous vehicle side" includes the manufacturer of the autonomous vehicle 5 (system developer, parts manufacturer, etc.), the operator of the remote monitoring device 6, etc. As described above, the first information generation unit 137 in this embodiment generates an overview diagram based on the detection results of the on-board sensors, in addition to the overview diagram based on the autonomous driving function (perception, judgment, control). Therefore, the negligence judgment unit 139 in this embodiment compares the first overview information based on the autonomous driving function (perception, judgment, control), the first overview information based on the detection results of the on-board sensors, and the second overview information.

[0048] In this embodiment, the negligence determination unit 139 compares, for each time based on time information, at least one of the following: presence or absence of a contact object, location of the contact object, location of the autonomous vehicle 5, traffic light color, movement trajectory, movement speed, road shape, compliance with traffic laws, presence or absence of surrounding objects, and location of surrounding objects. "Traffic light color" includes both the display color for the autonomous vehicle 5 and the display color for the other party. "Movement trajectory" and "movement speed" include both those of the autonomous vehicle 5 and those of the other party. "Compliance with traffic laws" includes not only those determined by referring to traffic management information such as whether or not the vehicle is stopped at a red light, but also those for which traffic management information is not available, such as whether or not the vehicle is waiting for a vehicle traveling on a priority road to pass.

[0049] For example, suppose a traffic accident occurs in which the vehicle makes contact with an object (pedestrian, parked vehicle, etc.) located on the shoulder of the road along the driving route. Then, suppose that, as shown in Figure 5, a first overview information based on the autonomous driving function, a first overview information based on the detection results of the on-board sensors, and a second overview information are generated at a certain time t. In this case, the negligence determination unit 139 compares the presence or absence of the object of contact and determines that the object is not depicted in the first overview information based on the autonomous driving function (recognition, judgment, control) (there was a problem with the function of recognizing objects, etc.). As a result, the negligence determination unit 139 will determine that the autonomous vehicle was at fault.

[0050] If the negligence determination unit 139 determines, after comparing the first overview information and the second overview information for each item at each time point, that the autonomous vehicle was not at fault, the calculation unit 13 terminates processing. If the second information generation unit 138 executes the second information generation process S8 before the first information generation process S7, the negligence determination unit 139 may execute the negligence determination process S9 after the first information generation process S7.

[0051] It should be noted that traffic accidents can also occur between autonomous vehicles. Therefore, in the event of a traffic incident involving two autonomous vehicles, the fault determination unit 139 will determine which autonomous vehicle was at fault.

[0052] • Data generation process S3, data generation unit 133 In the fault determination process S9, if the fault determination unit 139 determines that the autonomous vehicle was at fault, the process moves to the data generation process S3. In the data generation process S3, the data generation unit 133 generates reconstruction data of the traffic situation before the traffic accident based on time information, location information, function information, and detection results. As described above, the calculation unit 13 according to this embodiment includes a fault determination unit 139. Therefore, the data generation unit 133 according to this embodiment generates reconstruction data when the fault determination unit 139 determines that the autonomous vehicle was at fault. The fault determination unit 139 integrates first and second overview information for multiple time points to generate reconstruction data in video format, for example, as shown in Figure 6. That is, the reconstruction data is data showing the temporal changes of the positions of the autonomous vehicle 5, the other party involved in the accident, and other objects at and around the location where the traffic accident occurred, at least prior to the time the traffic accident occurred. As described above, the function information includes the vehicle log. Therefore, the reconstructed data also shows the planned route that the autonomous vehicle 5 was scheduled to travel, the communication status, and the status of the autonomous vehicle 5 and the other party.

[0053] Furthermore, as described above, the second acquisition unit 132 according to this embodiment also acquires traffic management information. Therefore, the data generation unit 133 according to this embodiment generates reproduced data taking into account the traffic management information. That is, the data generation unit 133 generates reproduced data that includes changes over time in the lighting color of traffic signals, changes over time in the opening and closing of barriers, changes over time in the display content of electronic display boards, and so on.

[0054] • Function identification process S4, function identification unit 134 The function identification unit 134 analyzes the reconstructed data to identify the problematic autonomous driving function among the multiple autonomous driving functions. As described above, the first acquisition unit 131 in this embodiment also acquires at least one of the vehicle log and the monitoring log. Therefore, the function identification unit 134 in this embodiment further analyzes at least one of the vehicle log and the monitoring log to identify the problematic autonomous driving function. Specifically, while reconstructing the data, the function identification unit 134 analyzes whether a situation corresponding to the occurrence of negligence, as shown in Figure 7, has occurred. If, for example, the analysis reveals that the object that caused the accident was not recognized, the function identification unit 134 identifies that there was a problem with the object recognition function among the autonomous driving functions of the autonomous driving vehicle 5. Also, if it is found that the settings for the driving environment conditions were inappropriate, the function identification unit 134 identifies that there was a problem with the remote monitoring function among the autonomous driving functions.

[0055] Output processing S10, output processing unit 140 After the function identification process S4, the process moves to the output process S10. In the output process S10, the output processing unit 140 controls the output unit. As a result, the output unit outputs information related to the identified automatic driving function. As described above, the output unit in this embodiment is also the communication unit 11. Therefore, the communication unit 11 transmits the information related to the identified automatic driving function to other systems 7, such as a display terminal (not shown).

[0056] Furthermore, in the output processing S10 according to this embodiment, the output processing unit 140 controls the output unit so that the second output unit outputs the reconstructed data. As described above, the second output unit according to this embodiment is also the communication unit 11. For this reason, the communication unit 11 transmits the reconstructed data to other systems 7, display terminals (not shown), etc. The output processing unit 140 may be configured to output the reconstructed data in its original format, or it may be configured to output it in a predetermined format. The predetermined format may include, for example, the format of police traffic accident records. By converting the data to a predetermined format in this way, the burden of record creation work at the receiving organization (e.g., the police) can be reduced.

[0057] [Variations of accident analysis device 1 (accident analysis system 100)] Disclosure 1 is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Furthermore, embodiments obtained by appropriately combining the technical means disclosed in the embodiments of Disclosure 2 described later are also included in the technical scope of Disclosure 1.

[0058] For example, the calculation unit 13 according to the above embodiment included a first information generation unit 137, a second information generation unit 138, and a fault determination unit 139. However, these functional blocks 136 to 138 may be provided by other calculation units or other devices different from the accident analysis device 1. In that case, the calculation unit 13 does not need to include these functional blocks 136 to 138.

[0059] Furthermore, in the above embodiment, the communication unit 11 also served as the output unit and the second output unit, but at least one of the output unit and the second output unit may be composed of a device different from the communication unit 11.

[0060] Furthermore, the accident analysis program may be recorded on one or more computer-readable recording media, rather than temporarily. These recording media may or may not be provided to each unit. In the latter case, the accident analysis program may be supplied to each unit via any wired or wireless transmission medium.

[0061] Furthermore, some or all of the functions of each control block 131 to 140 provided by the above-mentioned arithmetic unit 13 may be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each control block 131 to 140 are formed is also included in the scope of Disclosure 1. In addition, it is also possible to realize the functions of each control block 131 to 140 by, for example, a quantum computer.

[0062] [Effects and Effects of Accident Analysis Device 1 (Accident Analysis System 100)] The accident analysis device 1 (accident analysis system 100) described above generates highly accurate reconstructed data by incorporating objective information such as the detection results of the infrastructure sensor 3 into the data generation unit 133. Then, the function identification unit 134 analyzes the reconstructed data to objectively identify where in the autonomous vehicle a problem occurs that would result in a traffic accident like the one described in the reconstructed data. Therefore, the accident analysis device 1 makes it possible to specifically identify the cause of a traffic accident involving an autonomous vehicle without a driver. As a result, it becomes easy to determine which of the multiple businesses involved in autonomous driving should bear responsibility for negligence.

[0063] <Disclosure 2> Next, we will describe the second embodiment of disclosure.

[0064] [Background, challenges, solutions, and principles of solutions] Prior to describing the embodiments of Disclosure 2, we will explain the background of Disclosure 2, the problems of the prior art, the means for solving the problems, and the principle of solving them by the means.

[0065] 〔background〕 When a traffic accident involving a vehicle occurs, the cause of the accident must be identified, and the percentage of fault between the vehicle and the other party must be calculated. Currently, the calculation of fault percentages is done by insurance companies interviewing the parties involved in the accident and comparing their findings with past cases.

[0066] 〔assignment〕 However, conventional calculation methods are prone to subjective interpretation by insurance company employees. This can lead to problems such as the percentage of fault being calculated without a sufficient factual assessment, or the percentage of fault being calculated based on findings that differ from the facts.

[0067] [Solution] To solve the above problems, a fault ratio estimation system according to one aspect of the present disclosure includes: a first acquisition unit that acquires time information indicating the time, location information indicating the position of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition unit that acquires detection results prior to the time indicated by the time information from infrastructure sensors that are arranged along the driving route of the autonomous vehicle and detect traffic conditions; an information generation unit that generates overview information showing an overview of traffic conditions from a predetermined time before the traffic accident occurred, based on the time information, location information, function information, and detection results corresponding to the time when a traffic accident involving the autonomous vehicle occurred; a ratio estimation unit that inputs the overview information to a trained model constructed to output a numerical value when information regarding traffic conditions before the traffic accident is input, and acquires the numerical value output by the trained model as the fault ratio of the autonomous vehicle; and an output unit that outputs the fault ratio.

[0068] [Solution Principle] The fault ratio estimation system configured in this way generates overview information based on objective information such as the detection results of infrastructure sensors. The fault ratio estimation system then inputs the overview information into a trained model, and obtains the numerical value output by the trained model as the fault ratio of the autonomous vehicle. In this way, the fault ratio estimation system described in Disclosure 2 can calculate a highly objective fault ratio that does not involve the subjectivity of insurance company employees.

[0069] [Disclosure Embodiment 2] Next, an embodiment of Disclosure 2 will be described. For the sake of clarity, components having the same function as those described in the embodiment of Disclosure 1 will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0070] [Outline configuration of the fault ratio estimation system 100A] As shown in Figure 8, the fault ratio estimation system 100A includes a detection result database 2 similar to that of the accident analysis system 100 described in Disclosure 1, as well as a fault ratio estimation device 1A and an accident case database 8. Furthermore, the fault ratio estimation system 100A is connected via a network N to at least one infrastructure sensor 3, at least one traffic management device 4, and other systems 7, similar to those of the accident analysis system 100 described in Disclosure 1, as well as at least one vehicle 5A. Note that at least one of the detection result database 2 and the accident case database 8 does not have to be included in the fault ratio estimation system 100A. Also, at least one of the detection result database 2 and the accident case database 8 may be configured integrally with the fault ratio estimation device 1A. Also, at least one of the at least one traffic management device 4, at least one infrastructure sensor 3, at least one vehicle 5A, and other systems 7 may be included in the fault ratio estimation system 100A. Also, the fault ratio estimation system 100A may be incorporated into either the vehicle 5A or the other systems 7.

[0071] (Accident Case Database 8) Accident Case Database 8 stores past traffic accident cases (details of the accident, results of fault determination, fault percentages, etc.). Furthermore, whenever a new traffic accident occurs, Accident Case Database 8 stores its details, the results of the fault determination made at that time, the calculated fault percentages, etc., as a new case.

[0072] (Vehicle 5A) Vehicle 5A may be an autonomous driving vehicle 5 in the accident analysis system 100 related to Disclosure 1, a connected car, or a conventional vehicle that does not have an autonomous driving function.

[0073] (Fault Ratio Estimation Device 1A) As shown in Figure 9, the fault ratio estimation device 1A includes a communication unit 11 similar to that of the accident analysis device 1 described in Disclosure 1, as well as a storage unit 12A and a calculation unit 13A.

[0074] (Storage unit 12) The memory unit 12A stores the fault ratio estimation program 122. The fault ratio estimation program 122 is a program that causes the computer to function as a fault ratio estimation device 1A. The memory unit 12 also stores a trained model 123. The trained model 123 is configured to output a numerical value (ratio) when traffic information is input. "Traffic information" refers to information about the traffic situation before the traffic accident occurred. Traffic information includes overview information, which will be described later. The trained model 123 in this embodiment is configured to output a numerical value and the reason for outputting that numerical value when traffic information is input. The memory unit 12A may also be configured to store various calculation results of the arithmetic unit 13A. The memory unit 12A in this embodiment is composed of semiconductor memory, a hard disk drive, etc.

[0075] (Arithmetic unit 13A) The calculation unit 13A includes a first acquisition unit 131, a second acquisition unit 132, an accident occurrence determination unit 135, and a table creation unit 136, similar to the calculation unit 13 described in Disclosure 1, as well as an information generation unit 137A and a ratio estimation unit 141. The calculation unit 13A according to this embodiment further includes a model update unit 142 and an output processing unit 140A. The calculation unit 13A according to this embodiment is composed of a processor and memory. In other words, the fault ratio estimation device 1A is composed of a computer. Therefore, the functions of each control block 131 to 142 are realized by the calculation unit 13A executing various processes in the flow shown in Figure 10 according to the fault ratio estimation program 122 stored in the storage unit 12A.

[0076] • Information generation process 7A, Information generation unit 137A If the accident occurrence determination unit 135 determines that a traffic accident involving an autonomous vehicle has occurred, the process moves to information generation process S7A. In information generation process S7, the information generation unit 137A generates overview information that shows the traffic situation from the time of the accident to a predetermined time prior, based on time information, location information, function information, and detection results. The information generation unit 137A generates the overview information in the same manner as the first information generation unit 137 and the second information generation unit 138 generate the first and second overview information.

[0077] • Percentage estimation process S11, percentage estimation unit 141 After the information generation process 7A, the process moves to the ratio estimation process S11. In the ratio estimation process S11, the ratio estimation unit 141 inputs overview information into the trained model 123 and obtains the numerical value output by the trained model 123 as the fault ratio of the autonomous vehicle. As described above, the trained model 123 in this embodiment is configured to output a numerical value and the reason for outputting that numerical value when traffic information is input. Therefore, the ratio estimation unit 141 in this embodiment inputs overview information into the trained model and obtains the numerical value and reason output by the trained model as the fault ratio of the autonomous vehicle and the reason for that fault ratio. The reason output by the trained model 123 is used to verify whether the numerical value was appropriate or not.

[0078] Output processing S10A, output processing unit 140A After the ratio estimation process S11, the process moves to output processing S10A. In output processing S10A, the output processing unit 140A controls the output unit. As a result, the output unit outputs the fault ratio. As described above, the output unit in this embodiment is also the communication unit 11. Therefore, the communication unit 11 transmits the fault ratio to other systems 7, such as a display terminal (not shown). Also, as described above, the ratio estimation unit 141 in this embodiment obtains the reason for the fault ratio. Therefore, in output processing S10A in this embodiment, the output unit further outputs the reason. The fault ratio and reason output by the output unit can be used in a use case such as the one shown in Figure 11, in the manner shown in the figure. Although not shown in Figure 11, the fault ratio and reason output by the output unit can also be used to gather information about the accident situation from the parties involved in the accident.

[0079] • Model update section 142 Model update process S12 is performed as needed. In model update process S12, the model update unit 142 updates the trained model 123 based on overview information, fault ratio, reason, and evaluation results for the fault ratio and reason. The "evaluation results" may be made by a person, or they may be the output of an evaluation device configured to evaluate the fault ratio based on the reason.

[0080] [Example of the fault ratio estimation device 1A (fault ratio estimation system 100A)] Disclosure 2 is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Furthermore, embodiments obtained by appropriately combining the technical means disclosed in the embodiments of Disclosure 1 are also included in the technical scope of Disclosure 2.

[0081] For example, the calculation unit 13A according to the above embodiment included an information generation unit 137A and a model update unit 142. However, at least one of these functional blocks 137A and 142 may be provided by another calculation unit or another device different from the fault ratio estimation device 1. In that case, the calculation unit 13A does not need to include at least one of these functional blocks 137A and 142.

[0082] Furthermore, in the above embodiment, the communication unit 11 also served as the output unit and the second output unit, but at least one of the output unit and the second output unit may be composed of a device different from the communication unit 11.

[0083] Furthermore, the fault ratio estimation program may be recorded on one or more computer-readable recording media, rather than temporarily. Each unit may or may not have such recording media. In the latter case, the fault ratio estimation program may be supplied to each unit via any wired or wireless transmission medium.

[0084] Furthermore, some or all of the functions of each control block 131 to 142 provided by the above-mentioned arithmetic unit may be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each control block 131 to 142 are formed is also included in the scope of Disclosure 2. In addition, it is also possible to realize the functions of each control block 131 to 142 by, for example, a quantum computer.

[0085] [Effects and Effects of the Negligence Ratio Estimation Device 1A (Negligence Ratio Estimation System 100A)] As described above, the fault ratio estimation device 1A (fault ratio estimation system 100A) uses an information generation unit 137A to generate overview information based on objective information such as the detection results of infrastructure sensors. The ratio estimation unit 141 then inputs the overview information into a trained model, and the numerical value output by the trained model is obtained as the fault ratio of vehicle 5A. Therefore, the fault ratio estimation device 1A (fault ratio estimation system 100A) can calculate a highly objective fault ratio that does not involve the subjectivity of insurance company employees.

[0086] <Summary> An accident analysis system according to embodiment 1 of the present invention comprises: a first acquisition unit that acquires time information indicating the time, location information indicating the position of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition unit that acquires detection results from infrastructure sensors arranged along the driving path of the autonomous vehicle and detecting traffic conditions; a data generation unit that generates reconstruction data of traffic conditions prior to the time of the traffic accident based on the time information, location information, function information, and detection results corresponding to the time a traffic accident involving the autonomous vehicle occurred; a function identification unit that analyzes the reconstruction data to identify the problematic autonomous driving function among the plurality of autonomous driving functions; and an output unit that outputs information relating to the identified autonomous driving function.

[0087] The accident analysis system according to aspect 2 of the present invention further comprises, in aspect 1 above, a first information generation unit that generates first overview information showing an overview of the traffic situation from a predetermined time before the occurrence of the traffic accident based on recorded time information, location information, and function information; a second information generation unit that generates second overview information showing an overview of the traffic situation from a predetermined time before the occurrence of the traffic accident based on the time information, location information, and detection results; and a fault determination unit that compares the first overview information and the second overview information and determines whether or not the autonomous vehicle was at fault, wherein the data generation unit may be configured to generate the reconstruction data when the fault determination unit determines that the autonomous vehicle was at fault.

[0088] In the accident analysis system according to embodiment 3 of the present invention, in embodiment 2 described above, the negligence determination unit may be configured to compare, for each time based on the time information, at least one of the following: presence or absence of a contact object, location of the contact object, location of the autonomous vehicle, color of the traffic light, movement trajectory, movement speed, road shape, compliance with traffic laws, presence or absence of surrounding objects, and location of surrounding objects.

[0089] The accident analysis system according to aspect 4 of the present invention may be configured such that, in any of aspects 1 to 3 described above, the second acquisition unit further acquires traffic management information output by a traffic management device on the driving route, and the data generation unit generates the reconstruction data taking into account the traffic management information.

[0090] The accident analysis system according to embodiment 5 of the present invention may be configured such that, in any of embodiments 1 to 4 above, the first acquisition unit further acquires at least one of a vehicle log showing various operations performed by the autonomous vehicle and a monitoring log showing various operations performed by the remote monitoring device of the autonomous vehicle, and the function identification unit further analyzes at least one of the vehicle log and the monitoring log to identify the autonomous driving function that had the problem.

[0091] The accident analysis system according to aspect 6 of the present invention may further include a second output unit that outputs the reproduction data, as in any of the above aspects 1 to 5.

[0092] The accident analysis program according to aspect 7 of the present invention is configured to cause a computer to perform the following: a first acquisition process to acquire time information indicating the time, location information indicating the position of the autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition process to acquire detection results prior to the time indicated by the time information from infrastructure sensors that are placed along the driving route of the autonomous vehicle and detect traffic conditions; a data generation process to generate reconstruction data of the traffic conditions prior to the time of the traffic accident based on the time information, location information, function information, and detection results corresponding to the time when a traffic accident involving the autonomous vehicle occurred; a function identification process to analyze the reconstruction data and identify the autonomous driving function that had a problem among the plurality of autonomous driving functions; and an output process to output information relating to the identified autonomous driving function.

[0093] An accident analysis method according to aspect 8 of the present invention is a method comprising: a first acquisition step in which a computer acquires time information indicating a time, location information indicating the position of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition step in which the computer acquires detection results prior to the time indicated by the time information from infrastructure sensors that are placed along the driving route of the autonomous vehicle and detect traffic conditions; a data generation step in which the computer generates reconstruction data of the traffic conditions prior to the time of the traffic accident based on the time information, location information, function information, and detection results corresponding to the time when a traffic accident involving the autonomous vehicle occurred; a function identification step in which the computer analyzes the reconstruction data to identify the autonomous driving function that had a problem among the plurality of autonomous driving functions; and an output step in which the computer outputs information relating to the identified autonomous driving function.

[0094] A fault ratio estimation system according to aspect 9 of the present invention comprises: a first acquisition unit that acquires time information indicating a time, location information indicating the position of an autonomous vehicle, and function information relating to a plurality of autonomous driving functions of the autonomous vehicle; a second acquisition unit that acquires detection results prior to the time indicated by the time information from infrastructure sensors arranged along the autonomous vehicle's travel route and detecting traffic conditions; an information generation unit that generates overview information showing an overview of traffic conditions from a predetermined time before the traffic accident occurred, based on the time information, location information, function information, and detection results corresponding to the time a traffic accident involving the autonomous vehicle occurred; a ratio estimation unit that inputs the overview information to a trained model constructed to output a numerical value when traffic information relating to traffic conditions before the traffic accident is input, and acquires the numerical value output by the trained model as the fault ratio of the autonomous vehicle; and an output unit that outputs the fault ratio.

[0095] The fault ratio estimation system according to aspect 10 of the present invention may be configured such that, in aspect 9 above, the ratio estimation unit inputs the overview information into a trained model which is constructed to output a numerical value and the reason for outputting the numerical value when the traffic information is input, and obtains the numerical value and reason output by the trained model as the fault ratio of the autonomous vehicle and the reason for that fault ratio, and the output unit further outputs the reason.

[0096] The fault ratio estimation system according to aspect 11 of the present invention may further include a model update unit that updates the trained model based on the overview information, the fault ratio, the reason, and the evaluation results for the fault ratio and the reason, as described in aspect 10 above. [Explanation of Symbols]

[0097] 100 Accident Analysis System 1 Accident analysis device 11. Communication section (output section, second output section) 12, 12A storage section 121 Accident Analysis Program 13 Arithmetic section 131 First Acquisition Department 132 Second Acquisition Department 133 Data Generation Unit 135 Accident Occurrence Judgment Department 134 Function identification part 136 Table Creation Section 137 First Information Generation Department 138 Second information generation section 139 Negligence Judgment Division 140 Output Processing Unit 2. Detection Result Database 3. Infrastructure Sensors 4 Traffic management device 5. Autonomous vehicles 6 Remote monitoring device 7 Other systems 100A Negligence Ratio Estimation System 1A Negligence Ratio Estimation Device 12A storage section 122 Negligence Ratio Estimation Program 123 Pre-trained models 13A calculation section 137A Information generation section 140A Output Processing Unit 5A Vehicle 8. Accident Case Database

Claims

1. A first acquisition unit acquires time information indicating the time, location information indicating the position of the autonomous vehicle, and functional information relating to multiple autonomous driving functions of the autonomous vehicle. A second acquisition unit acquires detection results from infrastructure sensors that are placed along the driving path of the autonomous vehicle and detect traffic conditions. A data generation unit generates data recreating the traffic conditions prior to the time of the traffic accident, based on the time information, location information, function information, and detection results corresponding to the time the traffic accident involving the autonomous vehicle occurred. A function identification unit analyzes the aforementioned reproduction data to identify the problematic autonomous driving function among the multiple autonomous driving functions, An output unit that outputs information regarding identified autonomous driving functions, Equipped with, Accident analysis system.

2. A first information generation unit generates first overview information that shows the traffic situation from a predetermined time before the traffic accident occurred, based on the aforementioned time information, location information, and function information. A second information generation unit generates second overview information that shows an overview of the traffic situation from a predetermined time before the traffic accident occurred, based on the aforementioned time information, location information, and detection results. A fault determination unit compares the first overhead information and the second overhead information to determine whether or not the autonomous vehicle was at fault, Furthermore, The data generation unit generates the reconstruction data when the fault determination unit determines that the autonomous vehicle was at fault. The accident analysis system according to claim 1.

3. The negligence determination unit compares, for each time interval based on the time information, at least one of the following: the presence or absence of a contact object, the location of the contact object, the location of the autonomous vehicle, the color of the traffic light, the movement trajectory, the movement speed, the road shape, compliance with traffic laws, the presence or absence of surrounding objects, and the location of surrounding objects. The accident analysis system according to claim 2.

4. The second acquisition unit further acquires traffic management information output by traffic management devices along the travel route. The data generation unit generates the reproduced data by taking into account the traffic management information. The accident analysis system according to claim 1.

5. The first acquisition unit further acquires at least one of a vehicle log showing various operations performed by the autonomous vehicle and a monitoring log showing various operations performed by the remote monitoring device of the autonomous vehicle. The function identification unit further analyzes at least one of the vehicle log and the monitoring log to identify the problematic autonomous driving function. The accident analysis system according to claim 1.

6. The system further includes a second output unit that outputs the aforementioned reproduction data. The accident analysis system according to claim 1.

7. On the computer, A first acquisition process that acquires time information indicating the time, location information indicating the position of the autonomous vehicle, and functional information relating to multiple autonomous driving functions of the autonomous vehicle, A second acquisition process that acquires detection results prior to the time indicated by the time information from infrastructure sensors placed along the driving route of the autonomous vehicle to detect traffic conditions, A data generation process that generates data recreating the traffic conditions prior to the time of the traffic accident, based on the time information, location information, function information, and detection results corresponding to the time the traffic accident involving the autonomous vehicle occurred, The function identification process involves analyzing the aforementioned reproduction data to identify the problematic autonomous driving function among the multiple autonomous driving functions, Output processing that outputs information about identified autonomous driving functions, To execute Accident analysis program.

8. A first acquisition step involves the computer acquiring time information indicating the time, location information indicating the position of the autonomous vehicle, and functional information relating to multiple autonomous driving functions of the autonomous vehicle. A second acquisition step in which the computer acquires detection results prior to the time indicated by the time information from infrastructure sensors placed along the driving route of the autonomous vehicle to detect traffic conditions, A data generation step in which a computer generates data recreating the traffic conditions prior to the time of the traffic accident, based on the time information, location information, function information, and detection results corresponding to the time the traffic accident involving the autonomous vehicle occurred, A function identification step in which the computer analyzes the reproduction data to identify the problematic autonomous driving function among the multiple autonomous driving functions, The computer outputs information about the identified autonomous driving function in an output step, including, Accident analysis method.

9. A first acquisition unit acquires time information indicating the time, location information indicating the position of the autonomous vehicle, and functional information relating to multiple autonomous driving functions of the autonomous vehicle. A second acquisition unit that acquires detection results prior to the time indicated by the time information from infrastructure sensors placed along the driving route of the autonomous vehicle to detect traffic conditions, An information generation unit generates overview information that shows an overview of the traffic situation from a predetermined time before the traffic accident occurred, based on the time information, location information, function information, and detection results corresponding to the time when the traffic accident involving the autonomous vehicle occurred, A ratio estimation unit inputs the overview information into a trained model that is constructed to output a numerical value when traffic information regarding the traffic situation before the traffic accident occurs is input, and obtains the numerical value output by the trained model as the fault ratio of the autonomous vehicle. An output unit that outputs the aforementioned percentage of fault, Equipped with, A system for estimating the percentage of fault.

10. The ratio estimation unit inputs the overview information into a trained model that is configured to output a numerical value and the reason for outputting that numerical value when the traffic information is input, and obtains the numerical value and reason output by the trained model as the fault ratio of the autonomous vehicle and the reason for that fault ratio. The output unit further outputs the reason. The fault ratio estimation system according to claim 9.

11. The system further includes a model update unit that updates the trained model based on the overview information, the percentage of fault, the reason, and the evaluation results for the percentage of fault and the reason. The fault ratio estimation system according to claim 10.

Images