Control unit

Abstract

To provide a technology that enables stable automatic driving control.SOLUTION: A control unit (100) includes an event detection unit (110) and a control rule modification unit (120). The event detection unit (110) determines whether or not an event that triggers a change in a control rule during automatic driving of a vehicle is detected while the vehicle is driving automatically using a first control rule based on machine learning. When the triggering event is detected by the event detection unit (110), a control rule changing unit (120) changes the control rule during automatic driving of the vehicle to a second control rule according to the triggering event.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a control device, a control method, and a program. [Background technology]

[0002] In recent years, research and development into autonomous driving control has been progressing. For example, the following Patent Document 1 discloses a technology for performing hazard prediction and the like based on a knowledge base that stores logical expressions generated using a known supervised machine learning method, and utilizing the knowledge base for autonomous driving control of a vehicle. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2016-91039 A Summary of the Invention [Problem to be solved by the invention]

[0004] Models constructed by machine learning (machine learning models) change depending on the learning data input, and each machine learning model may have its own unique characteristics. Such machine learning models are basically preferable in that they enable optimal control to be executed based on the learning data given when the model was constructed. On the other hand, appropriate measures may not always be executed for input data (events) that deviate significantly from the learning data given when the model was constructed. For this reason, control based on machine learning models may not be preferable in situations where stable control is required.

[0005] One example of a problem to be solved by the present invention is to provide a technique that enables stable automatic driving control. [Means for solving the problem]

[0006] The invention described in claim 1 is an event detection unit that determines whether or not an event that triggers a change in a control rule during autonomous driving of the vehicle is detected while the vehicle is autonomously driven using a first control rule based on machine learning; a control rule changing unit that changes a control rule during autonomous driving of the vehicle to a second control rule corresponding to the triggering event when the triggering event is detected by the event detection unit; and The control device is provided with:

[0007] This disclosure includes: 1. A computer-implemented control method comprising: determining whether an event that triggers a change in a control rule during autonomous driving of the vehicle is detected while the vehicle is autonomously driven using a first control rule based on machine learning; When the triggering event is detected by the event detection unit, changing a control rule during autonomous driving of the vehicle to a second control rule corresponding to the triggering event; The present invention also includes a control method comprising:

[0008] This disclosure includes: Computer, an event detection means for determining whether or not an event that triggers a change in a control rule for autonomous driving of the vehicle has been detected while the vehicle is autonomously driven using a first control rule based on machine learning; and a control rule changing means for changing a control rule for autonomous driving of the vehicle to a second control rule corresponding to the triggering event when the triggering event is detected by the event detection unit; The program to function as a [Brief description of the drawings]

[0009] The above objects, as well as other objects, features and advantages, will become more apparent from the following preferred embodiments and the accompanying drawings.

[0010] [Figure 1] FIG. 1 is a diagram for explaining an overview of a control device according to the present invention. [Diagram 2] 2 is a block diagram conceptually showing a functional configuration of a control device in the first embodiment. FIG. [Diagram 3] FIG. 2 is a diagram illustrating an example of a hardware configuration of a control device according to the first embodiment. [Figure 4] 4 is a flowchart illustrating a flow of processing executed by the control device of the first embodiment. [Diagram 5] 11 is a diagram illustrating an example of information associating a predetermined event with a second control rule. FIG. [Figure 6] FIG. 11 is a block diagram conceptually showing a functional configuration of a control device in a second embodiment. [Figure 7] FIG. 11 is a diagram illustrating an example of a hardware configuration of a control device according to a second embodiment. [Figure 8] FIG. 11 is a sequence diagram illustrating a flow of processing executed by a control device of a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] [Summary] FIG. 1 is a diagram for explaining an overview of a control device 100 according to the present invention. In the example of FIG. 1, the control device 100 is a device (for example, an ECU (Electronic Control Unit)) mounted on a vehicle V. The control device 100 can change a first control rule that changes (optimizes) by machine learning and a second control rule that is a fixed rule not based on machine learning. When the control device 100 detects a predetermined event that triggers a change in the control rule for automatic driving while the vehicle is being driven automatically using the first control rule based on machine learning, as will be described in detail later, the control device 100 changes the control rule to a second control rule that is a fixed rule according to the event. Note that information on the predetermined event that triggers a change in the control rule for automatic driving can be obtained from an output from a sensor device 300 mounted on the vehicle V or an external device 500. The external device 500 is, for example, a device similar to the control device 100 mounted on another vehicle (not shown) or a device for road-to-vehicle communication provided along a road.

[0012] Hereinafter, the embodiments of the present invention will be described with reference to the drawings. In all drawings, the same components are given the same reference numerals, and the description will be omitted as appropriate. Furthermore, unless otherwise specified, each block in the block diagram represents a functional configuration, not a hardware configuration.

[0013] [First embodiment] 2 is a block diagram conceptually showing the functional configuration of the control device 100 in the first embodiment. As shown in FIG. 2, the control device 100 in this embodiment has an event detection unit 110 and a control rule change unit 120.

[0014] The event detection unit 110 determines whether or not a predetermined event is detected while the vehicle is autonomously driven using the first control rule based on machine learning. The predetermined event is an event that triggers a change in the control rule during autonomous driving of the vehicle. The predetermined event can also be expressed as an event that indicates the timing to interrupt the control of autonomous driving using the first control rule based on machine learning. A specific example of the predetermined event will be described later. The event detection unit 110 can detect the predetermined event based on the result of analyzing the output from various sensor devices 300 mounted on the vehicle. The event detection unit 110 may also detect the predetermined event via an external device (a control device mounted on another vehicle in the vicinity, or a device for road-to-vehicle communication installed along the road) by communicating with the external device. The event detection unit 110 may also detect the predetermined event based on dynamic information included in the map data for autonomous driving used during autonomous driving of the vehicle V. Here, the map data for autonomous driving is, for example, map data called a "dynamic map" and is data including conventional map information (static information) and information that changes in real time (dynamic information). The dynamic information includes information that may change in a relatively short span (for example, in units of seconds), such as ITS (Intelligent Transport Systems) look-ahead information (surrounding vehicles, pedestrian information, traffic signal information, etc.), and information that may change in a slightly short span (for example, in units of minutes), such as accident information, traffic congestion information, and narrow-area weather information (also called "quasi-dynamic information"). The static information includes information that may change in a relatively long span (for example, in units of months), such as road surface information, lane information, and three-dimensional structures, and information that may change in a slightly longer span (for example, in units of hours), such as traffic regulation information, road construction information, and wide-area weather information (also called "quasi-static information"). In the present invention, the quasi-static information may be classified into the category of dynamic information. Note that the static information and dynamic information included in the map data for automatic driving are not limited to the examples given here. The map data for automatic driving may include various information that can be used for automatic driving control of a vehicle.

[0015] When an event is detected by the event detection unit 110 while the vehicle is being driven autonomously using a first control rule based on machine learning, the control rule change unit 120 changes the control rule for the autonomous driving of the vehicle to a second control rule, which is a fixed rule that corresponds to the detected event.

[0016] The first control rule used by each vehicle during autonomous driving is a rule based on machine learning as described above. Therefore, the operation of the vehicle when the first control rule is used may have unique characteristics according to the learning results of the given learning data. When the first control rule is used to control the autonomous driving, the unique characteristics of the first control rule may cause unpredictable unstable operation. In this regard, in the present embodiment, when a predetermined event is detected while autonomous driving is being performed using the first control rule based on machine learning, the control rule for autonomous driving is changed to a second control rule, which is a fixed rule according to the detected event. As a result, when a predetermined event is detected, the operation of the vehicle during autonomous driving is controlled according to the fixed rule, so that unpredictable unstable operation can be suppressed.

[0017] In addition, in a place where a plurality of vehicles having the control device 100 are gathered, each vehicle can perform automatic driving using a fixed second control rule instead of a first control rule based on a machine learning result, and as a result, the movement of each vehicle can be controlled, and the traffic environment can be improved. For example, consider a case where there is an obstacle in one lane on a three-lane road, and each vehicle traveling in that lane needs to change lanes to avoid the obstacle. In this case, if each vehicle operates according to the first control rule having characteristics unique to each vehicle, a problem may occur in which vehicles concentrate in one of the remaining two lanes, causing or worsening congestion. In such a case, the control device 100 of the present embodiment switches the control rule of automatic driving of each vehicle to a fixed rule according to the detected event (in this case, for example, a rule such as "move to a lane different from the lane in which the vehicle in front moved as an obstacle avoidance action") and controls each vehicle, thereby distributing the vehicles to each of the remaining two lanes and minimizing the amount of congestion.

[0018] The control device 100 of this embodiment will be described in further detail below.

[0019] [Hardware configuration] Each functional component of the control device 100 may be realized by hardware that realizes each functional component (e.g., a hardwired electronic circuit, etc.), or may be realized by a combination of hardware and software (e.g., a combination of an electronic circuit and a program that controls it, etc.). Below, a further description will be given of the case where each functional component of the control device 100 is realized by a combination of hardware and software.

[0020] 3 is a diagram illustrating an example of a hardware configuration of the control device 100 of the first embodiment. A computer 200 is a computer that realizes the control device 100. For example, the computer 200 is an ECU (Electronic Control Unit) that can control the operation of a vehicle during autonomous driving. The computer 200 may be a computer designed specifically for realizing the control device 100, or may be a general-purpose computer.

[0021] The computer 200 has a bus 202, a processor 204, a memory 206, a storage device 208, an input / output interface 210, and a network interface 212. The bus 202 is a data transmission path for the processor 204, the memory 206, the storage device 208, the input / output interface 210, and the network interface 212 to transmit and receive data to and from each other. However, the method of connecting the processor 204 and the like to each other is not limited to bus connection. The processor 204 is an arithmetic processing device realized by using a microprocessor or the like. The memory 206 is a main storage device realized by using a RAM (Random Access Memory) or the like. The storage device 208 is an auxiliary storage device realized by using a ROM (Read Only Memory), a flash memory, or the like.

[0022] The input / output interface 210 is an interface for connecting the computer 200 to peripheral devices. Various analog signals and digital signals used for controlling a vehicle are input to or output from the computer 200 via the input / output interface 210. Here, the input / output interface 210 appropriately includes an A / D converter that converts an analog input signal into a digital signal, a D / A converter that converts a digital output signal into an analog signal, and the like.

[0023] For example, in Fig. 3, a sensor device 300 and a drive circuit 400 used for controlling a vehicle are connected to the input / output interface 210. The sensor device 300 is a LIDAR (Light Detection and Ranging), a millimeter wave radar, a sonar, a camera, or the like. Although not shown, a plurality of sensor devices 300 can be connected to the computer 200 via the input / output interface 210. The drive circuit 400 is a circuit for driving various mechanisms such as the gears, engine, and steering of the vehicle. The control device 100 can control the operation of the vehicle during autonomous driving by controlling the operation of the drive circuit 400.

[0024] The network interface 212 is an interface for connecting the computer 200 to a communication network. This communication network is, for example, a controller area network (CAN), a local area network (LAN), or a wide area network (WAN). The network interface 212 may connect to the communication network wirelessly or by wire. The computer 200 communicates with the control device 502 and the road-to-vehicle communication device 504 of another vehicle via a wireless LAN or the like, and can obtain information related to events used in the processing of the control device 100 from these devices.

[0025] The storage device 208 stores program modules for implementing each functional component of the control device 100. The processor 204 reads the program modules into the memory 206 and executes them to implement the functions of the control device 100. The storage device 208 may also store map data for autonomous driving that is used during autonomous driving of the vehicle V.

[0026] [Processing flow] The flow of processing executed by the control device 100 of this embodiment will be roughly described with reference to Fig. 4. Fig. 4 is a flowchart illustrating the flow of processing executed by the control device 100 of the first embodiment.

[0027] First, when the vehicle driving mode is switched to the autonomous driving mode (S102: YES), the event detection unit 110 is activated and starts a process of monitoring a predetermined event (S104). After that, when the event detection unit 110 detects a predetermined event (S104: YES), the event detection unit 110 notifies the control rule change unit 120 that the predetermined event has been detected (S106).

[0028] The control rule change unit 120 specifies the second control rule corresponding to the event notified in the process of S106 (S108). The control rule change unit 120 can specify the second control rule corresponding to the event detected by the event detection unit 110, for example, by using a table as shown in FIG. 5. The table illustrated in FIG. 5 stores identification information of the event and identification information of the second control rule to be applied in response to the detection of the event in association with each other. For example, the control rule change unit 120 can specify the second control rule by acquiring the identifier of the event detected in the process of S106 from the event detection unit 110 and referring to the table of FIG. 5 based on the identifier of the event. Then, the control rule change unit 120 transfers an instruction to apply the second control rule read in the process of S108 to an ECU that controls automatic driving or the like (S110). As a result, the operation of the vehicle during automatic driving is controlled based on the second control rule. Note that the explanation here is merely an example, and the operation of the control rule change unit 120 is not limited to using the table illustrated in FIG. 5. For example, while FIG. 5 shows an example in which a different second control rule is associated with each event, the present invention is not limited to this, and the same second rule may be associated with a plurality of events.

[0029] The operation will be described in more detail below by giving some concrete examples.

[0030] <First concrete example> In this specific example, a case will be described in which the event detection unit 110 detects "the occurrence of an abnormality in the sensor device 300" as a predetermined event. Here, the abnormality in the sensor device 300 means an abnormality in the output signal from the sensor device 300 or a communication failure between the sensor device 300 and the control device 100 due to contamination of the optical system (lens, etc.), internal failure, disturbance in the sensing environment (sunlight, rain, fog, snow, oncoming vehicle lights, etc.), or detection of a signal indicating an error or an unexpected signal.

[0031] A case will be described in which the event detection unit 110 detects a predetermined event based on "dynamic information included in the map data for automatic driving". As described above, the map data for automatic driving is a digital map incorporating not only static information (map information such as road surface information, lane information, and three-dimensional structures) but also dynamic information (accident information, traffic congestion information, weather information, pedestrian information, traffic light information, etc.). Specifically, the dynamic information included in the map data for automatic driving means the dynamic information. This dynamic information is distributed to the vehicle V from a server that manages accident information, etc., for example. The vehicle V that receives the accident information, etc. stores the accident information, etc. in an area indicating the dynamic information in the map data for automatic driving. The event detection unit 110 can detect an event by referring to the dynamic information included in the map data for automatic driving used during automatic driving of the vehicle V.

[0032] The event detection unit 110 monitors a signal line connected to the sensor device 300 via the input / output interface 210, measures the strength of a signal output from the sensor device 300, and analyzes the contents of the signal. The event detection unit 110 then determines whether the strength of the measured signal is lower than a predetermined reference value, or whether the analyzed signal is an error signal or an unexpected signal. In this case, the predetermined reference value to be compared is stored in advance in, for example, the memory 206 or the storage device 208. When it is detected that the strength of the measured signal is equal to or lower than the predetermined reference value, or that the analyzed signal is an error signal or an unexpected signal, the event detection unit 110 notifies the control rule change unit 120 that "an abnormality has occurred in the sensor device 300". Based on the notification from the event detection unit 110, the control rule change unit 120 specifies a second control rule to be applied when "an abnormality has occurred in the sensor device 300". The second control rule in this case is not particularly limited, but may be, for example, a rule such as "stop the vehicle in a predetermined procedure (for example, turn on the hazard lights and control the brakes to gradually decelerate, etc.)".

[0033] If the strength of the output signal from the sensor device 300 is reduced due to an abnormality, the probability that the control processing unit of the automatic driving will erroneously recognize the situation around the vehicle increases, and the operation during automatic driving is likely to become unstable. When an abnormality in the sensor device 300 is detected, the automatic driving control is performed according to a second control rule in which a fixed control operation is defined, thereby preventing the operation during automatic driving from becoming unstable. Note that when an abnormality in the sensor device 300 is detected, the control device 100 may be configured to give up the control by automatic driving and transfer the control authority to the driver instead of changing to the second control rule.

[0034] <Second concrete example> In this specific example, a case will be described in which the event detection unit 110 detects, as a specified event, "an accident occurring in at least one of the multiple lanes while a vehicle is traveling on a road having multiple lanes on one side."

[0035] The event detection unit 110 can detect a predetermined event by acquiring information indicating the position of the accident vehicle (e.g., position coordinates on map information, information on the lane where the accident has occurred, etc.) via the control device 502 of the other vehicle or the road-to-vehicle communication device 504 installed along the road. When the control device 502 of the other vehicle detects the presence of the accident vehicle using various sensors mounted on the other vehicle, for example, the control device 502 of the other vehicle can generate information notifying the presence of the accident vehicle together with the position information of the accident vehicle (e.g., position coordinates on map information, information on the lane where the accident has occurred, etc.) and transmit this information by vehicle-to-vehicle communication. In this case, the event detection unit 110 can detect a predetermined event via the control device 502 of the other vehicle by performing vehicle-to-vehicle communication with the control device 502 of the other vehicle. In addition, the road-to-vehicle communication device 504 can collect information notifying the presence of the accident vehicle and the position information of the accident vehicle from the control device 502 of the other vehicle and broadcast the collected information within the jurisdiction area of ​​the road-to-vehicle communication device 504. In this case, the event detection unit 110 can detect a predetermined event via the road-to-vehicle communication device 504 by receiving information broadcast from the road-to-vehicle communication device 504. In addition, when the sensor device 300 mounted on the vehicle is a camera having an image sensor, the presence or absence of an accident vehicle (predetermined event) can be detected by analyzing image data generated by the camera. Similarly, when the sensor device 300 mounted on the vehicle is a LIDAR, the presence or absence of an accident vehicle (predetermined event) can be detected based on an image generated from point cloud data obtained by laser light scanning of the LIDAR. For example, the event detection unit 110 can determine the presence or absence of an accident vehicle in image data by using a CNN (Convolutional Neural Network) constructed using images of the accident vehicle as learning data.

[0036] When the event detection unit 110 detects the above-mentioned predetermined event, for example, the control rule change unit 120 changes the control rule used during autonomous driving to a default rule before the first control rule is constructed by machine learning. Here, the default rule is a control rule in an initial state before machine learning is performed, in other words, a control rule that does not have inherent characteristics. By controlling autonomous driving using such a default rule, it is possible to suppress unpredictable unstable behavior caused by inherent characteristics generated by machine learning.

[0037] As another example, if the purpose is to control the operation among a plurality of vehicles, when the event detection unit 110 detects the above-mentioned predetermined event, the control rule change unit 120 may change the control rule used during automatic driving to a rule (hereinafter, also referred to as a "common rule") that is commonly used among a plurality of control devices (the control device 100, the control device 502 of the other vehicle, and control devices mounted on other vehicles not shown). The common rule may be prepared as a rule common to the whole world, a country, a region, a vehicle model, or a vehicle manufacturer. These common rules are stored in advance in the memory 206 or the storage device 208 in a format as shown in FIG. 5. The common rule may also be stored in map data for automatic driving stored in the storage device 208 or a server device outside the vehicle V. By using a rule common to a plurality of control devices, the variation in the operation of each vehicle during automatic driving is reduced, and the operation of each vehicle can be controlled. A specific example of the common rule is not particularly limited, but may be "move to a lane different from the lane in which the vehicle in front has moved".

[0038] <Third Specific Example> In this specific example, a case will be described in which the event detection unit 110 detects "the presence of an obstacle (a fallen object, a depression or flooding in the road surface, etc.) on the lane on which the vehicle is traveling" as a predetermined event.

[0039] The event detection unit 110 can detect a predetermined event by acquiring information indicating the position of an obstacle (e.g., position coordinates on map information, information on the lane in which the obstacle exists, etc.) via the control device 502 of the other vehicle or the road-to-vehicle communication device 504 installed along the road. When the control device 502 of the other vehicle detects an obstacle using various sensors mounted on the other vehicle, for example, the control device 502 can generate information notifying the presence of the obstacle together with the position information of the obstacle (e.g., position coordinates on map information, information on the lane in which the obstacle exists, etc.) and transmit this information by vehicle-to-vehicle communication. In this case, the event detection unit 110 can detect a predetermined event via the control device 502 of the other vehicle by performing vehicle-to-vehicle communication with the control device 502 of the other vehicle. In addition, the road-to-vehicle communication device 504 can collect information notifying the presence of an obstacle and the position information of the obstacle from the control device 502 of the other vehicle and broadcast the collected information within the jurisdiction area of ​​the road-to-vehicle communication device 504. In this case, the event detection unit 110 can detect a predetermined event via the road-to-vehicle communication device 504 by receiving information broadcast from the road-to-vehicle communication device 504. Also, the event detection unit 110 can detect the presence or absence of an obstacle (predetermined event) using the sensor device 300 mounted on the vehicle. For example, the event detection unit 110 can recognize the shape of the road surface and obstacles on the road surface based on an image generated using an image sensor or a scanning result (distance image) by LIDAR.

[0040] In this case, as in the second specific example, the control rule change unit 120 can change the control rule to be used during autonomous driving to a default rule before the first control rule was constructed by machine learning or to a rule that is commonly used among multiple control devices.

[0041] <Fourth Specific Example> In this specific example, a case will be described in which the event detection unit 110 detects, as a specified event, "the receipt of information indicating an action that the vehicle should take as a second control rule from the control device 502 or the road-vehicle communication device 504 of another vehicle."

[0042] As an example, the control device 502 of the other vehicle first detects the presence or absence of an obstacle on the road surface or an accident vehicle. The control device 502 of the other vehicle can detect the obstacle on the road surface or the accident vehicle based on, for example, an output from a sensor device provided in the other vehicle. The control device 502 of the other vehicle can also acquire information on the obstacle on the road surface or the accident vehicle from the road-to-vehicle communication device 504. When an obstacle on the road surface or an accident vehicle is detected, the control device 502 of the other vehicle controls the other vehicle to avoid the obstacle or the accident vehicle, and generates information indicating the action to be taken by the following vehicle. Then, the control device 502 of the other vehicle assigns, for example, a dedicated identifier to the generated information and transmits it to the following vehicle via the communication device. As another example, the road-to-vehicle communication device 504 may detect the presence or absence of an obstacle on the road surface or the accident vehicle based on an output from a sensor device provided in the other vehicle. The road-to-vehicle communication device 504 collects, for example, information (sensor information) on obstacles on the road surface and accident vehicles from surrounding vehicles, and uses the collected information to determine the action to be taken by each vehicle in the target area. The road-to-vehicle communication device 504 broadcasts the information indicating the determined action within the target area after, for example, assigning a dedicated identifier to the information. As a specific example of an instruction, when an obstacle is detected in the center lane of a three-lane road, the control device 502 or the road-to-vehicle communication device 504 of the other vehicle can transmit an instruction to the effect that a vehicle whose license plate has an odd-numbered last digit should move to the left lane, and a vehicle whose license plate has an even-numbered last digit should move to the right lane. When the event detection unit 110 receives information indicating the action to be taken by the vehicle from the control device 502 or the road-to-vehicle communication device 504 of the other vehicle, the event detection unit 110 uses the information as a second control rule to control the action of the vehicle during automatic driving. The event detection unit 110 can determine that the information received from the other vehicle is "information indicating an action to be taken by the vehicle itself" based on an identifier assigned to the information. In this specific example, the operation during autonomous driving is controlled based on instructions transmitted from the control device 502 and the road-to-vehicle communication device 504 of the other vehicle. This prevents unstable operation caused by inherent characteristics of machine learning.

[0043] [Second embodiment] In the first embodiment, an example has been shown in which a common rule is used among a plurality of control devices (the control device 100, the control device 502 of the other vehicle, and control devices mounted on other vehicles (not shown)). The control device 100 of this embodiment further includes a configuration for updating this common rule.

[0044] [Functional configuration] Fig. 6 is a block diagram conceptually showing a functional configuration of the control device 100 in the second embodiment. As shown in Fig. 6, the control device 100 in this embodiment further includes an update information acquisition unit 130 and a common rule update unit 140. In addition, as shown in Fig. 7, in this embodiment, a server device 506 that generates information for updating the common rule is communicably connected to the control device 100.

[0045] The update information acquisition unit 130 acquires update information of the common rule from the server device 506. As an example, the server device 506 learns using the result of the automatic driving control based on the common rule (for example, a change in the degree of congestion) as a reward, and updates the common rule. Specifically, the server device 506 acquires information indicating how each vehicle operated according to the common rule and information indicating a change in the degree of congestion from sensor devices and the like installed in each vehicle and around the road, and evaluates the current common rule. Then, the server device 506 updates the common rule based on the evaluation result. For example, assume that a reward of "the degree of congestion worsened by more than an allowable value" was obtained as a result of performing "action A" defined in the common rule. In this case, information is generated to update the common rule so as to give a certain penalty or a penalty according to the degree of deterioration to "action A" and lower the priority of selecting action A. In addition to the example given here, a configuration can be adopted in which an administrator manually inputs update information of the common rule into the server device 506 and distributes it to the control device of each vehicle. The server device 506 distributes the update information of the common rule thus generated to each vehicle. The server device 506 may be configured to distribute a new common rule updated using the update information of the common rule to the control device of each vehicle. The update information of the common rule distributed from the server device 506 may be distributed to the control device 100 of each vehicle via the road-to-vehicle communication device 504.

[0046] The common rule update unit 140 uses the update information for the common rule acquired by the update information acquisition unit 130 to update the common rule stored in the memory 206 or the storage device 208 .

[0047] [Hardware configuration] 7 is a diagram illustrating an example of the hardware configuration of the control device 100 of the second embodiment. In this embodiment, the storage device 208 stores program modules for realizing the functions of the update information acquisition unit 130 and the common rule update unit 140. The processor 204 reads these program modules into the memory 206 and executes them to respectively realize the functions of the update information acquisition unit 130 and the common rule update unit 140. In this embodiment, a server device 506 is connected via a network interface 212.

[0048] [Example of operation] The flow of processing executed by the control device 100 of the second embodiment will be described with reference to Fig. 8. Fig. 8 is a sequence diagram illustrating the flow of processing executed by the control device 100 of the second embodiment.

[0049] The server device 506 collects log information of the automatic driving control based on the common rule (for example, information indicating the operation and time of the common rule selected during automatic driving) from a plurality of control devices mounted on each vehicle (S202). The server device 506 also acquires information indicating a change in the degree of congestion together with time information from the road-to-vehicle communication device 504 installed along the road (S204). Then, the server device 506 generates update information of the common rule based on the operation selected by each vehicle according to the current common rule and the result (reward) associated therewith (S206). The server device 506 can identify the correspondence relationship of the information collected in the processes of S202 and S204 based on the time information. Then, the server device 506 distributes the update information of the common rule generated in the process of S206 to the control device of each vehicle (S208). The server device 506 may once transmit the update information of the common rule to the road-to-vehicle communication device 504 and distribute it to each vehicle via the road-to-vehicle communication device 504.

[0050] In each vehicle, the update information of the common rule distributed in S208 is received by the update information acquisition unit 130. The common rule update unit 140 updates the current common rule stored in the memory 206 or the storage device 208 based on the update information of the common rule acquired by the update information acquisition unit 130 (S210).

[0051] As described above, according to this embodiment, it is possible to optimize the common rule stored in each vehicle by using update information for the common rule distributed from the server device 506.

[0052] Although the embodiment and examples have been described above with reference to the drawings, these are merely examples of the present invention, and various configurations other than those described above can also be adopted.

[0053] This application claims priority based on Japanese Patent Application No. 2017-138957, filed on July 18, 2017, the disclosure of which is incorporated herein by reference in its entirety.

Claims

[Claim 1] an event detection unit that determines whether or not an event that triggers a change in a control rule during autonomous driving of the vehicle is detected while the vehicle is autonomously driven using a first control rule based on machine learning; a control rule changing unit that changes a control rule during autonomous driving of the vehicle to a second control rule corresponding to the triggering event when the triggering event is detected by the event detection unit; and A control device comprising:

Images