Automated driving system

Abstract

To provide a technique that can ensure a capability of explanation of automated driving control of a vehicle while reducing a data capacity when storing log data related to the automated driving control.SOLUTION: An automated driving system is mounted on a vehicle and performs automated driving of the vehicle using a machine learning model. When the automated driving is performed, the automated driving system is configured to execute: processing of acquiring log data related to the automated driving; processing of acquiring a current state of an operator of the vehicle; processing of acquiring an ideal state that is a state of the operator required by the automated driving system; and log data storage processing of storing the log data in one or more storage devices. When the current state of the operator matches the ideal state, the log data storage processing includes compressing or deleting target data that is part of the log data.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to an autonomous driving system that uses a machine learning model to autonomously drive a vehicle. [Background technology]

[0002] Patent Document 1 discloses a vehicle driving recorder that records monitoring image data of the surroundings of a vehicle. According to the vehicle driving recorder, image data obtained by a camera mounted on the vehicle capturing images of the surroundings of the vehicle is stored. However, not all of the image data is stored, but the driver's line of sight and the range capturing moving objects around the vehicle are recorded as surrounding situation monitoring image data.

[0003] In addition to the above-mentioned Patent Document 1, the following Patent Documents 2 and 3 can be mentioned as examples of documents showing the state of the art in the technical field related to the present disclosure. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2020-106921 A [Patent Document 2] Patent Publication No. 2021-051604 [Patent Document 3] JP 2009-015789 A Summary of the Invention [Problem to be solved by the invention]

[0005] As disclosed in Patent Document 1, devices that record information while a vehicle is traveling, such as drive recorders, are known, and these can be used for analyzing the cause of an accident. The same can be said about systems that perform automatic driving using machine learning models, in that it is required to store information in case verification is required later. In other words, it is important to store data related to automatic driving so that it is possible to verify later how automatic driving using a machine learning model was performed.

[0006] However, if all data related to autonomous driving is stored, the capacity of the storage device may be exhausted. Therefore, the challenge is how to ensure the explainability of autonomous driving control while conserving the capacity of the storage device.

[0007] One objective of the present disclosure is to provide a technology that enables the storage of log data related to autonomous driving control of a vehicle while reducing data volume and ensuring the explainability of the autonomous driving control. [Means for solving the problem]

[0008] The present disclosure provides an autonomous driving system that is installed in a vehicle and drives the vehicle autonomously using a machine learning model. The autonomous driving system includes one or more processors and one or more memory devices. When the one or more processors perform the automatic operation, A process of acquiring log data related to autonomous driving; Obtaining a current state of the vehicle operator; A process of acquiring an ideal state, which is the state of the operator desired by the autonomous driving system; a log data storage process for storing the log data in one or more storage devices; The apparatus is configured to execute the following steps: The log data storage process includes compressing or deleting target data of a portion of the log data when the operator's current state matches the ideal state. Effect of the Invention

[0009] According to the technology disclosed herein, the autonomous driving system compresses or deletes part of the log data depending on the conditions, and stores the compressed or deleted part of the log data in a storage device. Specifically, when the operator's current state matches the ideal state, the compressed or deleted part of the data is considered to mean that the operator is sufficiently monitoring the autonomous driving. In other words, by compressing or deleting the data when it is confirmed that the operator is sufficiently monitoring the autonomous driving, it is possible to reduce the amount of data to be stored while ensuring the explainability of the autonomous driving control. [Brief description of the drawings]

[0010] [Figure 1] FIG. 2 is a diagram showing a configuration example related to automatic driving control of a vehicle according to the present embodiment. [Diagram 2] FIG. 1 is a diagram illustrating an example of a configuration of an automatic driving system according to an embodiment of the present invention. [Diagram 3] 4 is a flowchart showing an example of processing performed by the autonomous driving system according to the present embodiment. [Figure 4] 10 is a flowchart showing a modified example of the processing performed by the autonomous driving system according to the present embodiment. [Diagram 5] 11 is a conceptual diagram showing an example of setting of log storage priority according to the embodiment. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] 1.Autonomous driving of vehicles The autonomous driving system according to the present embodiment is mounted on a vehicle and performs autonomous driving control of the vehicle. FIG. 1 is a block diagram showing a configuration example related to the autonomous driving control of the vehicle 1. Autonomous driving refers to automatically performing at least one of steering, acceleration, and deceleration of the vehicle 1 without relying on human operation. Autonomous driving control is a concept that includes not only fully autonomous driving control, but also risk avoidance control, lane keep assist control, and the like.

[0012] The vehicle 1 includes a sensor group 10, a recognition unit 20, a planner 30, a control amount calculation unit 40, and a traveling device 50.

[0013] The sensor group 10 includes a recognition sensor 11 used to recognize the situation around the vehicle 1. Examples of the recognition sensor 11 include a camera, a LIDAR (Laser Imaging Detection and Ranging), a radar, and the like. The camera may include a plurality of cameras, such as a front camera that captures an image in front of the vehicle 1, a right side camera that captures an image on the right side of the vehicle 1, a left side camera that captures an image on the left side, and a rear camera that captures an image on the rear side. The sensor group 10 may further include a state sensor 12 that detects a state of the vehicle 1, a position sensor 13 that detects a position of the vehicle 1, and the like. Examples of the state sensor 12 include a speed sensor, an acceleration sensor, a yaw rate sensor, a steering angle sensor, and the like. Examples of the position sensor 13 include a GNSS (Global Navigation Satellite System) sensor.

[0014] The sensor detection information SEN is information obtained by the sensor group 10. For example, the sensor detection information SEN includes images captured by a camera. As another example, the sensor detection information SEN may include point cloud information obtained by a LIDAR. The sensor detection information SEN may include vehicle state information indicating the state of the vehicle 1. The sensor detection information SEN may include position information indicating the position of the vehicle 1.

[0015] The recognition unit 20 receives the sensor detection information SEN. The recognition unit 20 recognizes the situation around the vehicle 1 based on the information obtained by the recognition sensor 11. For example, the recognition unit 20 recognizes objects around the vehicle 1. Examples of objects include pedestrians, other vehicles (preceding vehicles, parked vehicles, etc.), white lines, road structures (e.g., guardrails, curbs), fallen objects, traffic lights, intersections, signs, etc. The recognition result information RES indicates the recognition result by the recognition unit 20. For example, the recognition result information RES includes object information indicating the relative position and relative speed of the object with respect to the vehicle 1.

[0016] The planner 30 receives the recognition result information RES from the recognition unit 20. The planner 30 may also receive vehicle state information, position information, and pre-generated map information. The map information may be high-precision three-dimensional map information. The planner 30 generates a driving plan for the vehicle 1 based on the received information. The driving plan may be for reaching a pre-set destination, or for avoiding risks. Examples of the driving plan include maintaining the current driving lane, changing lanes, overtaking, turning right or left, steering, accelerating, decelerating, and stopping. Furthermore, the planner 30 generates a target trajectory TRJ required for the vehicle 1 to travel according to the driving plan. The target trajectory TRJ includes a target position and a target speed.

[0017] The control amount calculation unit 40 receives the target trajectory TRJ from the planning unit 30. The control amount calculation unit 40 calculates a control amount CON required for the vehicle 1 to follow the target trajectory TRJ. It can also be said that the control amount CON is a control amount required to reduce the deviation between the vehicle 1 and the target trajectory TRJ. The control amount CON includes at least one of a steering control amount, a drive control amount, and a braking control amount. Examples of the steering control amount include a target steering angle, a target torque, a target motor angle, a target motor drive current, etc. Examples of the drive control amount include a target speed, a target acceleration, etc. Examples of the braking control amount include a target speed, a target deceleration, etc.

[0018] The traveling device 50 includes a steering device 51, a driving device 52, and a braking device 53. The steering device 51 steers the wheels. For example, the steering device 51 includes an electric power steering (EPS) device. The driving device 52 is a power source that generates a driving force. Examples of the driving device 52 include an engine, an electric motor, and an in-wheel motor. The braking device 53 generates a braking force. The traveling device 50 receives a control amount CON from the control amount calculation unit 40. The traveling device 50 operates the steering device 51, the driving device 52, and the braking device 53 according to the steering control amount, the driving control amount, and the braking control amount, respectively. As a result, the vehicle 1 travels so as to follow the target trajectory TRJ.

[0019] The recognition unit 20 includes at least one of a rule-based model and a machine learning model. The rule-based model performs recognition processing based on a group of predetermined rules. Examples of the machine learning model include a Neural Network (NN), a Support Vector Machine (SVM), a regression model, a decision tree model, and the like. The NN may be a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), or a combination thereof. The type of each layer, the number of layers, and the number of nodes in the NN are arbitrary. The machine learning model is generated in advance through machine learning. The recognition unit 20 performs recognition processing by inputting the sensor detection information SEN to the model. The recognition result information RES is output from the model, or is generated based on the output from the model.

[0020] Similarly, the planner 30 includes at least one of a rule-based model and a machine learning model. The planner 30 performs planning processing by inputting the recognition result information RES to the model. The target trajectory TRJ is output from the model or is generated based on the output from the model.

[0021] Similarly, the control amount calculation unit 40 includes at least one of a rule-based model and a machine learning model. The control amount calculation unit 40 performs a control amount calculation process by inputting a target trajectory TRJ to the model. The control amount CON is output from the model or is generated based on the output from the model.

[0022] Two or more of the recognition unit 20, the planner 30, and the control amount calculation unit 40 may be integrally configured. All of the recognition unit 20, the planner 30, and the control amount calculation unit 40 may be integrally configured (end-to-end configuration). For example, the recognition unit 20 and the planner 30 may be integrally configured by a NN that outputs a target trajectory TRJ from the sensor detection information SEN. Even in the case of an integral configuration, intermediate products such as recognition result information RES and target trajectory TRJ may be output. For example, when the recognition unit 20 and the planner 30 are integrally configured by a NN, the recognition result information RES may be the output of an intermediate layer of the NN.

[0023] The recognition unit 20, the planning unit 30, and the control amount calculation unit 40 constitute an "automatic driving control unit" that controls the automatic driving of the vehicle 1. In this embodiment, a machine learning model is used in at least a part of the automatic driving control unit. That is, at least one of the recognition unit 20, the planning unit 30, and the control amount calculation unit 40 includes a machine learning model. The automatic driving control unit performs at least a part of the automatic driving control of the vehicle 1 by using the machine learning model.

[0024] As described above, the automatic driving control unit of the automatic driving system performs automatic driving of the vehicle 1. However, even during execution of the automatic driving control, the automatic driving system does not control the vehicle 1 completely independently of human hands. Even while the automatic driving control of the vehicle 1 is being performed, an operator is on standby in the vehicle 1 so that the vehicle 1 can be operated and monitored.

[0025] The operator is, for example, a passenger or a driver who boards the vehicle 1. The operator performs operations related to the control of the vehicle 1 depending on the situation. For example, the operator monitors the situation of the automatic driving and the situation around the vehicle 1, and performs operations to intervene in the automatic driving control for at least one of steering, acceleration, and deceleration of the vehicle 1 when it is judged necessary. Alternatively, the operator may perform necessary operations in response to a request from the automatic driving system. For example, when there is a vehicle traveling at a low speed ahead of the vehicle 1, the automatic driving system may request the operator's decision on whether or not to overtake the vehicle ahead. In this case, the operator inputs an instruction to the automatic driving system regarding whether or not to overtake in response to a request from the automatic driving system.

[0026] In addition, as requests from the automatic driving system to the operator, a compulsory operation request and a voluntary operation request are assumed. An example of a compulsory operation request is a hands-on request. For example, when the recognition unit 20 cannot correctly recognize the lane markings, the automatic driving system requests the driver to take over driving, which corresponds to a compulsory operation request. If the operator does not respond to such a compulsory request, the vehicle 1 cannot continue traveling, and the automatic driving system may stop the vehicle 1. In contrast, as for a voluntary operation request, the vehicle 1 can continue traveling even if there is no response from the operator. An example of a voluntary operation request is a request for the operator's judgment on whether or not to overtake or change lanes. For example, suppose that the automatic driving system requests the operator's judgment on whether or not to overtake a vehicle ahead. In this case, even if there is no response from the operator, the automatic driving system only controls the vehicle 1 to continue traveling by following the vehicle ahead, and the vehicle 1 itself does not become unable to continue traveling.

[0027] As another example, the operator may be an operator waiting at a remote terminal outside the vehicle 1, and may operate and monitor the vehicle 1 remotely.

[0028] 2 is a conceptual diagram showing a configuration example of an autonomous driving system 100 according to this embodiment. The autonomous driving system 100 is mounted on a vehicle 1 and performs autonomous driving control of the vehicle 1. The autonomous driving system 100 has at least the functions of the autonomous driving control unit described above. The autonomous driving system 100 may further include a sensor group 10 and a traveling device 50.

[0029] The autonomous driving system 100 is configured to be able to communicate with the operator sensor 60 and the input device 70 via a wired network or a wireless network. The operator sensor 60 is a sensor that acquires the state of the operator. Examples of the operator sensor 60 include a driver monitor (camera) and a camera that captures images inside the vehicle. The input device 70 is a device through which the operator inputs operations. The input device 70 includes an accelerator pedal, a brake pedal, a steering wheel, a touch panel, buttons, and the like. For example, the operator inputs an instruction as to whether or not to overtake into the input device 70 in response to a request from the autonomous driving system 100.

[0030] The autonomous driving system 100 includes one or more processors 110 (hereinafter simply referred to as processors 110) and one or more memory devices 120 (hereinafter simply referred to as memory devices 120).

[0031] The processor 110 executes various processes. Examples of the processor 110 include a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), and a field-programmable gate array (FPGA).

[0032] The storage device 120 stores a program 130 and various information. Examples of the storage device 120 include a hard disk drive (HDD), a solid state drive (SSD), a volatile memory, and a non-volatile memory. The various information stored in the storage device 120 includes log data LOG related to the automatic driving control.

[0033] The program 130 is a computer program for controlling the vehicle 1, and is executed by the processor 110. Various processes by the autonomous driving system 100 may be realized by cooperation between the processor 110, which executes the program 130, and the storage device 120. The program 130 may be recorded in a computer-readable recording medium.

[0034] While the vehicle 1 is performing autonomous driving using a machine learning model, the processor 110 acquires log data LOG related to the autonomous driving and stores it in the storage device 120.

[0035] The log data LOG includes at least system data related to the behavior of the autonomous driving system 100. The system data includes data input to the autonomous driving control unit and data output from the autonomous driving control unit. The system data may include an output of an intermediate layer of the autonomous driving control unit. For example, the system data includes at least a part of the sensor detection information SEN, the recognition result information RES, the target trajectory TRJ, and the control amount CON. The system data may also include data indicating the reason for the judgment in the recognition process by the recognition unit 20, the reason for the judgment in the planning process by the planner 30, and the like. The stored system data is used for verifying the autonomous driving control using a machine learning model, etc.

[0036] In addition to the system data, the log data LOG may include operator data related to the state of the operator. For example, the operator data may include a facial image of the operator captured by a camera included in the operator sensor 60. As another example, the operator data may include the line of sight direction of the operator obtained by analyzing the facial image. As yet another example, the operator data may include an operation state of the input device 70 by the operator.

[0037] 2. Processing related to the storage of log data A process related to the storage of log data LOG by the processor 110 will be described. Fig. 3 is a flowchart showing an example of a process executed by the autonomous driving system 100, more specifically, the processor 110. The process according to the flowchart shown in Fig. 3 is executed repeatedly at a predetermined processing cycle, for example, while the autonomous driving control is being executed.

[0038] In step S110, the processor 110 acquires log data LOG related to the autonomous driving.

[0039] In step S120, the processor 110 acquires the current state of the operator. The current state of the operator is the state of the operator at the time when the log data LOG is acquired, and is acquired by at least one of the operator sensor 60 and the input device 70.

[0040] For example, the data indicating the current state of the operator is data of an image obtained by capturing an image of the operator by the camera of the operator sensor 60. Furthermore, the processor 110 may determine in which direction the face of the operator is facing by image analysis of the image capturing the operator, and may acquire the direction in which the operator is facing as the current state. Alternatively, data indicating what operation the operator inputted to the input device 70 may be acquired as the data indicating the current state. As another example of data acquired from the input device 70, the presence or absence of intervention in the automatic driving control by the operator may be acquired as the current state. Alternatively, the presence or absence of a response of the operator to an operation request from the automatic driving system 100 may be acquired as the current state.

[0041] In step S130, the processor 110 sets an ideal state. The ideal state is a state that the automatic driving system 100 requests the operator to perform some operation. For example, when the automatic driving system 100 requests the operator to perform some operation, the ideal state is set to be that the operator responds to the operation request. Alternatively, when the vehicle 1 is traveling steadily forward, the ideal state may be set to be that the operator monitors the road ahead. Alternatively, when the automatic driving system 100 performs a lane change, the ideal state may be set to be that the operator checks the adjacent lane, that is, that the operator looks toward the adjacent lane.

[0042] In step S140, processor 110 determines whether the current state of the operator matches the ideal state. If the current state of the operator matches the ideal state (step S140; Yes), the process proceeds to step S150. On the other hand, if the current state of the operator does not match the ideal state (step S140; No), the process proceeds to step S160.

[0043] Whether or not the current state of the operator matches the ideal state is determined by comparing the current state acquired in step S120 with the ideal state acquired in step S130. For example, if the ideal state is set to be an operator responding to an operation request and an operation responding to a request from the automatic driving system 100 is input to the input device 70, it is determined that the current state matches the ideal state. Alternatively, if the ideal state is set to be an operator monitoring the front and it is determined that the operator is facing forward through image analysis of an image obtained by capturing an image of the operator with the camera of the operator sensor 60, it is determined that the current state matches the ideal state.

[0044] In step S150, the processor 110 compresses or deletes a portion of the data in the log data LOG. Hereinafter, in this specification, the portion of the data in the log data LOG that is the target of compression or deletion is referred to as the target data. Compressing the target data means making the target data into coarse data. Deleting the target data means completely deleting the target data. Note that in step S150, all of the target data may be compressed or deleted, or a portion of the target data may be deleted and a portion of the target data may be compressed.

[0045] For example, the target data includes operator data. For example, if the ideal state is for the operator to monitor the front and the operator faces forward, it is not necessarily necessary to store the face image of the operator itself. By storing information that the current state of the operator matches the ideal state instead of the face image of the operator itself, the amount of stored data can be reduced.

[0046] In step S160, the processor 110 stores the log data LOG in the storage device 120. If it is determined in step S140 that the current state of the operator matches the ideal state, the log data LOG in which the target data has been compressed or deleted is stored. If it is determined in step S140 that the current state of the operator does not match the ideal state, the log data LOG acquired in step S110 is stored as is. Once the log data LOG is stored, the series of processes ends.

[0047] The effect of the series of processes will be described. As described above, the log data LOG is stored in the storage device 120 in preparation for cases where it becomes necessary to later verify how the autonomous driving control was performed. In other words, in order to fulfill the accountability for the autonomous driving, it is desirable to store as much of the data acquired as the log data LOG as possible. However, since the capacity of the storage device 120 is limited, there is a possibility that the capacity will be tight if all of the log data LOG is stored. Therefore, it is required to reduce the amount of data stored while ensuring the explainability for the autonomous driving.

[0048] According to the autonomous driving system 100 of the present embodiment, the log data LOG is stored in which the target data is compressed or deleted depending on the conditions. By compressing or deleting the target data, the amount of data to be stored can be reduced compared to the case where all the acquired log data LOG is stored, and the capacity of the storage device 120 can be saved.

[0049] Furthermore, the compression or deletion of the target data is not performed unconditionally, but is performed when the current state of the operator matches the ideal state. The current state of the operator matching the ideal state means that the operator is in a state required by the automated driving system 100, in other words, the operator is sufficiently monitoring the automated driving system 100. In such a case, the operator understands how the automated driving by the automated driving system 100 was performed, and it is considered that there is a low possibility that verification using the log data LOG is required. In such a case, the compression or deletion of the target data can ensure the explainability required for the log data LOG while reducing the data volume.

[0050] Note that the processing from step S140 to step S160 does not have to be performed simultaneously with the acquisition of the log data LOG. For example, while automatic driving is being performed, the processing from step S110 to step S130 is repeatedly executed at a predetermined control period, and all of the log data LOG is temporarily stored in the storage device 120. Then, at any timing thereafter, it is determined whether or not the current state of the operator at the time the log data LOG is acquired matches the ideal state, and if they match, the target data may be deleted from the temporarily stored log data LOG or compressed.

[0051] 3.Determining target data based on log storage priority The processor 110 may assign a log storage priority to each piece of data included in the log data LOG. Here, the log storage priority is a priority for storage, and a higher log storage priority indicates that the data should be stored with higher priority.

[0052] An example of processing when log storage priorities are assigned to data included in the log data LOG is shown in the flowchart of Fig. 4. The processing from step S110 to step S140 is omitted because it is similar to the flowchart of Fig. 3. However, in step S140, if it is determined that the current state of the operator matches the ideal state (step S140; Yes), the processing proceeds to step S151.

[0053] In step S151, the processor 110 acquires the remaining capacity of the storage device 120. More specifically, the processor 110 acquires the remaining capacity of an area of ​​the storage device 120 that has been secured for saving the log data LOG.

[0054] In step S152, the processor 110 assigns a log storage priority to each piece of data included in the log data LOG. The log storage priority is assigned so that it is high for data that is likely to be needed later for verification or the like, and low for data that is unlikely to be needed. Data that is likely to be needed later is data that indicates the status of a part that is not adequately monitored by the operator. The processor 110 can determine which ranges the operator is monitoring and which ranges he is not monitoring, based on the current state of the operator acquired in step S120.

[0055] FIG. 5 is a conceptual diagram for explaining an example of log storage priority. Here, in the case where the recognition sensor 11 includes a plurality of cameras, the log storage priority of the image captured by the camera of the recognition sensor 11 among the data included in the log data LOG will be considered. When the operator faces forward, there is a high possibility that the operator is sufficiently monitoring the range captured by the front camera. On the other hand, there is a high possibility that the operator is not sufficiently monitoring the range captured by the side camera or the rear camera. In such a case, the processor 110 lowers the log storage priority of the image captured by the front camera and raises the log storage priority of the image captured by the other cameras.

[0056] As another specific example, it is assumed that the recognition result information RES by the recognition unit 20 includes the recognition result for the right side and the recognition result for the left side of the vehicle 1. Then, it is assumed that a state in which the operator is looking at the right side is acquired in step S120 by image analysis of the image captured by the camera of the operator sensor 60. In this case, it is highly likely that the operator is monitoring the situation on the right side of the vehicle 1, but not the left side. Therefore, the data indicating the recognition result for the left side is given a higher log storage priority than the data indicating the recognition result for the right side.

[0057] When the log data LOG includes operator data, a log storage priority may be assigned to each piece of data included in the operator data. In this case as well, the log storage priority is assigned so that data that is likely to be required in a later verification or other situation has a high log storage priority, and data that is unlikely to be required has a low log storage priority.

[0058] An example of how to assign log storage priorities to data included in the operator data will be described below. For example, the log storage priorities may be assigned in descending order of log storage priority to data indicating the presence or absence of an operator's response to a compulsory operation request from the automated driving system 100, data indicating the presence or absence of an operator's response to a voluntary operation request from the automated driving system 100, and data indicating the direction of the operator's line of sight.

[0059] The operator's response to a compulsory operation request has a greater impact on the automatic driving of the vehicle 1 than the operator's response to a voluntary operation request. As described above, the vehicle 1 can continue traveling regardless of whether or not there is a response from the operator to a voluntary operation request, whereas if there is no response to a compulsory operation request, the vehicle 1 may not be able to continue traveling. Therefore, the presence or absence of an operator's response to a compulsory operation request is highly likely to be important in subsequent verification and other situations. Therefore, a high log storage priority is given to such data.

[0060] In addition, data showing the direction of the operator's gaze is not necessarily required even if a post-facto verification is required. This type of data is useful as a basis for determining whether the target data may be compressed or deleted, but does not directly affect the automatic driving control. For this reason, it is conceivable to set such a log storage priority.

[0061] Furthermore, a log storage priority may be assigned to data indicating the direction of the operator's line of sight. Examples of data indicating the direction of the operator's line of sight include the image itself obtained by the camera of the operator sensor 60 capturing an image of the operator's face, data indicating the angle of the operator's face, and data indicating only whether the operator is facing the direction required by the automatic driving system 100. These three data indicate the same content, but the latter data can be said to be more abstract. For these data, log storage priorities are assigned so that the more abstract the data, the higher the log storage priority. This is because the more abstract the data, the smaller the data volume, and therefore the less likely it is to overwhelm the capacity of the storage device 120 when saved.

[0062] In step S153, processor 110 determines target data based on the log save priority determined in step S152 and the remaining capacity of storage device 120 acquired in step S151. More specifically, processor 110 prioritizes data with a low log save priority as target data over data with a high log save priority. Processor 110 also selects target data such that the total amount of target data is small when the remaining capacity of storage device 120 is large, and the total amount of target data is large when the remaining capacity of storage device 120 is small.

[0063] Once the target data has been determined, the process proceeds to step S150 in the flowchart of FIG. 3, where the target data is compressed or deleted.

[0064] In this way, when a log storage priority is assigned to each data included in the log data LOG, it is more effective in ensuring the explainability of the log data LOG. Data on the status of a portion that is sufficiently monitored by an operator is less likely to be needed for later verification, etc. Therefore, by selecting target data from data with a low log storage priority, data to be saved from the log data LOG that is more likely to be needed later can be selected. In this way, while compressing or deleting the target data, important information is prevented from being deleted first, and the explainability and ease of verification of the log data LOG can be more effectively guaranteed. [Explanation of symbols]

[0065] 1 ... vehicle, 10 ... sensor group, 20 ... recognition unit, 30 ... planning unit, 40 ... control amount calculation unit, 50 ... driving device, 60 ... operator sensor, 70 ... input device, 100 ... automatic driving system, 110 ... processor, 120 ... storage device, CON ... control amount, RES ... recognition result information, SEN ... sensor detection information, TRJ ... target trajectory

Claims

1. An autonomous driving system that is mounted on a vehicle and uses a machine learning model to autonomously drive the vehicle, one or more processors; One or more storage devices; Equipped with When the one or more processors execute the automatic driving, A process of acquiring log data related to the autonomous driving; obtaining a current status of an operator of the vehicle; A process of acquiring an ideal state, which is a state of the operator desired by the automated driving system; a log data storage process for storing the log data in the one or more storage devices; configured to run The log data storage process includes compressing or deleting target data of a portion of the log data when the current state of the operator matches the ideal state. An automated driving system characterized by

2. The automated driving system according to claim 1, The log data includes system data related to a behavior of the autonomous driving system and operator data related to a state of the operator. An automated driving system characterized by

3. The automated driving system according to claim 2, The subject data includes the operator data. An automated driving system characterized by

4. The automatic driving system according to any one of claims 1 to 3, The one or more processors are further configured to execute a priority determination process for assigning a log storage priority to each data included in the log data; When the current state of the operator matches the ideal state, the log data storage process prioritizes data with a low log storage priority over data with a high log storage priority as the target data. An automated driving system characterized by

5. The automated driving system according to claim 4, The log data includes recognition data obtained by recognizing the surroundings of the vehicle, The priority determination process includes setting the log storage priority of the recognition data for an area monitored by the operator lower than the log storage priority of the recognition data for an area not monitored by the operator. An automated driving system characterized by

Images