JP2021067259A - Vehicle control system, vehicle control device, and vehicle learning device - Google Patents

Abstract

To provide a vehicle control system that allows reduction in the number of man-hours necessary for an expert in setting a relationship between a state of a vehicle and an action variable.SOLUTION: A CPU 72 sets a throttle opening degree command value and a retardation amount of ignition timing as an action by using relationship definition data DR on the basis of time-series data of an accelerator operation amount PA as a state, operates a throttle valve and an ignition device according to them, and acquires a torque, a torque command value and an acceleration at that time. The CPU 72 transmits the state, the action, the torque, the torque command value and the acceleration to a data analysis center 110. The data analysis center 110 updates the relationship definition data DR by rewarding according to whether the torque and the acceleration satisfy criteria on the basis of the data transmitted from multiple vehicles VC1, VC2, ....SELECTED DRAWING: Figure 1

Description

The present invention relates to a vehicle control system, a vehicle control device, and a vehicle learning device.

For example, Patent Document 1 below describes a control device that operates a throttle valve as an operation unit of an internal combustion engine mounted on a vehicle based on a value obtained by filtering the operation amount of the accelerator pedal.

Japanese Unexamined Patent Publication No. 2016-6327

By the way, in the above filter, it is necessary to set the operation amount of the throttle valve of the internal combustion engine mounted on the vehicle to an appropriate operation amount according to the operation amount of the accelerator pedal. However, it is necessary to spend a lot of man-hours. As described above, conventionally, a skilled person has spent a lot of man-hours to adapt the operation amount of the electronic device in the vehicle according to the state of the vehicle.

Hereinafter, means for solving the above problems and their actions and effects will be described.
1. 1. A storage device that stores relationship regulation data that defines the relationship between the state of the vehicle and an action variable that is a variable related to the operation of electronic devices in the vehicle, a first execution device mounted on the vehicle, and an in-vehicle device. A second execution device different from the above, the acquisition process of acquiring the detection value of the sensor for detecting the state of the vehicle, the operation process of operating the electronic device, and the detection acquired by the acquisition process. Based on the values, a reward calculation process that gives a larger reward than when the characteristics of the vehicle meet the criteria, a state of the vehicle based on the detected value acquired by the acquisition process, and an operation of the electronic device. The value of the action variable used in the above and the reward corresponding to the operation are input to a predetermined update mapping, and the update process for updating the related regulation data is performed on the first execution device and the first execution device. 2 Execution is performed in cooperation with the execution device, the first execution device executes at least the acquisition process and the operation process, and the operation process is the related regulation data updated by the update process and the state of the vehicle. The updated mapping includes the process of operating the electronic device in accordance with the above, and the updated mapping is updated to increase the expected return on the reward when the electronic device is operated according to the relevant provision data. The update process outputs data, and the update process obtains the states of the plurality of vehicles, the values of the action variables used for the operation of the electronic devices of the plurality of vehicles, and the reward corresponding to the operation. It is a vehicle control system executed by the second execution device as an input to the update mapping.

In the above configuration, by calculating the reward associated with the operation of the electronic device, it is possible to grasp what kind of reward can be obtained by the operation. Then, based on the reward, the relationship between the state of the vehicle and the behavior variable can be set to an appropriate relationship in the running of the vehicle by updating the relational regulation data by the update mapping according to the reinforcement learning. Therefore, when setting the relationship between the state of the vehicle and the behavior variable to an appropriate relationship in the running of the vehicle, the man-hours required for the expert can be reduced.

Further, in the above configuration, a single relational data based on the state, behavior variable and reward of a single vehicle is updated by updating a single relational data based on the state, behavior variable and reward of a plurality of vehicles. The update frequency can be increased as compared with the case of updating.

2. In the update process, the states of the plurality of vehicles belonging to the same group and the electronic devices of the plurality of vehicles belonging to the same group in order to update the related regulation data for each group in which the vehicles are classified. The second execution device includes the process of updating the relational regulation data belonging to the group by inputting the value of the action variable used for the operation and the reward corresponding to the operation into the update mapping, and the second execution device is the vehicle. 1 is the vehicle control system according to 1 above, which executes a specific process for specifying which of the plurality of groups the state, the value of the action variable, and the reward belong to.

The pair of the state of the vehicle and the value of the appropriate behavior variable may differ depending on the driving environment of the vehicle, the degree of wear, the driving preference of the user, and the like. Therefore, in the above configuration, by updating the relational regulation data for each group, a single relational regulation data is updated based on the vehicle state, the behavior variable, and the reward for all the vehicles equipped with the first execution device in the system. Compared to the case, it is possible to update the relational regulation data to more appropriate data for each group.

3. 3. The specific process is the vehicle control system according to the above 2, which includes a process of specifying which group the vehicle belongs to according to the position of the vehicle.
Since the situation in which the vehicle is placed differs depending on the region, the value of the action variable appropriate for the condition of the vehicle may differ depending on the region. Therefore, in the above configuration, by specifying the group based on the position information variable, it is possible to update the relational regulation data to the data suitable for each of the plurality of regions.

4. The vehicle control system according to 2 or 3 above, wherein the specific process includes a process of specifying which group the vehicle belongs to according to the degree of wear of the vehicle.
Since the responsiveness of the vehicle to the operation by the user differs depending on the degree of wear of the vehicle, the value of the action variable appropriate for the state of the vehicle may differ depending on the degree of wear of the vehicle. Therefore, in the above configuration, by specifying the group according to the degree of wear of the vehicle, it is possible to update to the appropriate relational regulation data according to the degree of wear of the vehicle.

5. The operation process includes a search process for executing an operation different from the operation for maximizing the expected return grasped from the relational regulation data, and the second execution device is the action variable to be executed as the search process. The vehicle according to any one of 1 to 4 above, which executes an instruction process for instructing different values between the first vehicle and the second vehicle, which are two of the plurality of vehicles. It is a control system.

In the above configuration, it is possible to control the information obtained from the search results of each of the plurality of vehicles by instructing the first vehicle and the second vehicle to execute different actions as a search.

6. A vehicle control device including the first execution device according to any one of 1 to 5 above.
7. A vehicle learning device including the second execution device according to any one of 1 to 5 above.

The figure which shows the structure of the control system for a vehicle which concerns on 1st Embodiment. The flow chart which shows the procedure of the process executed by the control device which concerns on the same embodiment. (A) and (b) are flow charts showing the procedure of processing executed by the system according to the same embodiment. The figure which illustrates the classification of the vehicle which concerns on this embodiment. (A) and (b) are flow charts showing the procedure of the process executed by the system which concerns on 2nd Embodiment. The figure which illustrates the setting of the action for the re-search concerning the same embodiment.

<First Embodiment>
Hereinafter, the first embodiment of the vehicle control system will be described with reference to the drawings.
FIG. 1 shows the configuration of the drive system and the control system of the vehicle VC1 according to the present embodiment.

As shown in FIG. 1, the intake passage 12 of the internal combustion engine 10 is provided with a throttle valve 14 and a fuel injection valve 16 in this order from the upstream side, and is injected from the air or fuel injection valve 16 sucked into the intake passage 12. The fuel is discharged into the combustion chamber 24 partitioned by the cylinder 20 and the piston 22 as the intake valve 18 is opened. In the combustion chamber 24, the air-fuel mixture is subjected to combustion with the spark discharge of the ignition device 26, and the energy generated by the combustion is converted into the rotational energy of the crankshaft 28 via the piston 22. To. The air-fuel mixture used for combustion is discharged to the exhaust passage 32 as exhaust gas when the exhaust valve 30 is opened. The exhaust passage 32 is provided with a catalyst 34 as an aftertreatment device for purifying the exhaust gas.

The input shaft 52 of the transmission 50 can be mechanically connected to the crankshaft 28 via a torque converter 40 provided with a lockup clutch 42. The transmission 50 is a device that makes the gear ratio variable, which is the ratio of the rotation speed of the input shaft 52 to the rotation speed of the output shaft 54. A drive wheel 60 is mechanically connected to the output shaft 54.

The control device 70 targets the internal combustion engine 10, and in order to control the torque and the exhaust component ratio, which are the controlled amounts thereof, the control device 70 controls the operation unit of the internal combustion engine 10 such as the throttle valve 14, the fuel injection valve 16, and the ignition device 26. Manipulate. Further, the control device 70 controls the torque converter 40 and operates the lockup clutch 42 in order to control the engaged state of the lockup clutch 42. Further, the control device 70 targets the transmission 50 as a control target, and operates the transmission 50 to control the gear ratio as the control amount thereof. Note that FIG. 1 shows the operation signals MS1 to MS5 of the throttle valve 14, the fuel injection valve 16, the ignition device 26, the lockup clutch 42, and the transmission 50, respectively.

The control device 70 controls the intake air amount Ga detected by the air flow meter 80, the opening degree of the throttle valve 14 (throttle opening degree TA) detected by the throttle sensor 82, and the crank angle sensor 84 for controlling the control amount. Refer to the output signal Scr of. Further, the control device 70 refers to the depression amount of the accelerator pedal 86 (accelerator operation amount PA) detected by the accelerator sensor 88 and the acceleration Gx in the front-rear direction of the vehicle VC1 detected by the acceleration sensor 90. Further, the control device 70 refers to the position data Pgps by the Global Positioning System (GPS92).

The control device 70 includes a CPU 72, a ROM 74, an electrically rewritable non-volatile memory (storage device 76), a communication device 77, and a peripheral circuit 78, which can communicate with each other via a local network 79. Here, the peripheral circuit 78 includes a circuit that generates a clock signal that defines the internal operation, a power supply circuit, a reset circuit, and the like.

The control program 74a and the learning subprogram 74b are stored in the ROM 74. On the other hand, the storage device 76 has the relational regulation data DR that defines the relationship between the accelerator operation amount PA, the command value of the throttle opening degree TA (throttle opening degree command value TA *), and the retard angle amount aop of the ignition device 26. It is remembered. Here, the retard angle amount aop is a retard angle amount with respect to a predetermined reference ignition timing, and the reference ignition timing is a timing on the retard side of the MBT ignition timing and the knock limit point. The MBT ignition timing is the ignition timing at which the maximum torque can be obtained (maximum torque ignition timing). The knock limit is an ignition timing advance limit that can keep knocking within an acceptable level under the best possible conditions when using a high octane fuel with a high knock limit. Further, the torque output mapping data DT is stored in the storage device 76. The torque output map defined by the torque output map data DT is a map that outputs the torque Trq by inputting the rotation speed NE, the filling efficiency η, and the ignition timing of the crankshaft 28.

The communication device 77 is a device for communicating with the data analysis center 110 via the network 100 outside the vehicle VC1.
The data analysis center 110 analyzes the data transmitted from the plurality of vehicles VC1, VC2, .... The data analysis center 110 includes a CPU 112, a ROM 114, an electrically rewritable non-volatile memory (storage device 116), peripheral circuits 118, and a communication device 117, which are made communicable by the local network 119. Is. The ROM 114 stores a learning main program 114a for updating the relational regulation data DR by reinforcement learning. Further, the storage device 116 stores the related regulation data DR.

FIG. 2 shows a procedure of processing executed by the control device 70 according to the present embodiment. The process shown in FIG. 2 is realized by the CPU 72 repeatedly executing the control program 74a stored in the ROM 74, for example, at a predetermined cycle. In the following, the step number of each process is indicated by a number prefixed with "S".

In the series of processes shown in FIG. 2, the CPU 72 first, as the state s, time-series data composed of six sampling values "PA (1), PA (2), ... PA (6)" of the accelerator operation amount PA. (S10). Here, each sampling value constituting the time series data is sampled at different timings from each other. In the present embodiment, time-series data is configured by six sampling values that are adjacent to each other in time series when sampling is performed at a constant sampling cycle.

Next, the CPU 72 sets the action a including the throttle opening degree command value TA * and the retard angle amount aop according to the state s acquired by the process of S10, in accordance with the policy π defined by the relational regulation data DR (S12).

In the present embodiment, the relational regulation data DR is data that defines the action value function Q and the policy π. In the present embodiment, the action value function Q is a table-type function showing the value of the expected return according to the eight-dimensional independent variables of the state s and the action a. Further, the policy π is predetermined while preferentially selecting the action a (greedy action) that is the largest among the action value functions Q in which the independent variable is given the state s when the state s is given. With the probability ε of, the rule for selecting other actions a is determined.

Next, the CPU 72 outputs the operation signal MS1 to the throttle valve 14 to operate the throttle opening degree TA based on the set throttle opening degree command value TA * and the retard angle amount aop, and operates the throttle opening degree TA, and at the same time, operates the operation signal MS3 to the ignition device 26. Is output to control the ignition timing (S14). Here, in the present embodiment, since the feedback control of the throttle opening degree TA to the throttle opening degree command value TA * is illustrated, even if the throttle opening degree command value TA * is the same value, the operation signal MS1 Can be different signals from each other. Further, for example, when a well-known knocking control (KCS) or the like is performed, the ignition timing is a value obtained by retarding the reference ignition timing with a retard angle amount aop and feedback-corrected by KCS. Here, the reference ignition timing is variably set by the CPU 72 according to the rotation speed NE of the crankshaft 28 and the filling efficiency η. The rotation speed NE is calculated by the CPU 72 based on the output signal Scr of the crank angle sensor 84. Further, the filling efficiency η is calculated by the CPU 72 based on the rotation speed NE and the intake air amount Ga.

Next, the CPU 72 acquires the torque Trq of the internal combustion engine 10, the torque command value Trq * for the internal combustion engine 10, and the acceleration Gx (S16). Here, the CPU 112 calculates the torque Trq by inputting the rotation speed NE, the filling efficiency η, and the ignition timing into the torque output map. Further, the CPU 72 sets the torque command value Trq * according to the accelerator operation amount PA.

Next, the CPU 72 determines whether or not the transient flag F is “1” (S18). When the transient flag F is "1", it indicates that it is in transient operation, and when it is "0", it indicates that it is not in transient operation. When the CPU 72 determines that the transient flag F is "0" (S18: NO), the CPU 72 determines whether or not the absolute value of the change amount ΔPA of the accelerator operation amount PA per unit time is equal to or greater than the predetermined amount ΔPAth. (S20). Here, the change amount ΔPA may be, for example, the difference between the latest accelerator operation amount PA at the execution timing of the process of S20 and the accelerator operation amount PA before the same timing by a unit time.

When the CPU 72 determines that the predetermined amount is ΔPAth or more (S20: YES), the CPU 72 substitutes “1” for the transient flag F (S22).
On the other hand, when the CPU 72 determines that the transient flag F is "1" (S18: YES), the CPU 72 determines whether or not a predetermined period has elapsed after the execution of the process of S22 (S24). Here, the predetermined period is a period until the state in which the absolute value of the change amount ΔPA of the accelerator operation amount PA per unit time is smaller than the predetermined amount ΔPAth and is equal to or less than the specified amount continues for a predetermined time. When the CPU 72 determines that the predetermined period has elapsed (S24: YES), the CPU 72 substitutes “0” for the transient flag F (S26).

When the processing of S22 and S26 is completed, or when a negative determination is made in the processing of S20 and S24, the CPU 72 shifts to the processing of S28. In the processing of S28, the CPU 72 includes the state s acquired in the processing of S10, the action a selected in the processing of S12, the torque Trq acquired in the processing of S16, the torque command value Trq *, the acceleration Gx, and the transient flag F. And the value of are stored in the storage device 76.

When the process of S28 is completed, the CPU 72 temporarily ends the series of processes shown in FIG.
FIG. 3 shows a processing procedure of reinforcement learning according to the present embodiment. The process shown in FIG. 3A is realized by the CPU 72 executing the learning subprogram 74b stored in the ROM 74 shown in FIG. Further, the process shown in FIG. 3B is realized by the CPU 112 executing the learning main program 114a stored in the ROM 114. Hereinafter, the processes shown in FIG. 3 will be described in chronological order.

In the series of processes shown in FIG. 3A, the CPU 72 first determines whether or not it is the end of the trip (S30). Here, the trip is a period of one time in which the traveling permission signal of the vehicle is in the ON state. In the present embodiment, the travel permission signal corresponds to an ignition signal.

When the CPU 72 determines that it is the end of the trip (S30: YES), the CPU 72 operates the communication device 77 and stores the identification information ID of the vehicle VC1, the mileage Lt, the position data Pgps, and the processing of S28. The data is transmitted (S32).

On the other hand, as shown in FIG. 3B, the CPU 112 receives the data transmitted by the process of S32 (S40). Then, the CPU 112 selects one period, that is, one episode in which the transient flag F of the received data is constant (S42). Each episode is a period from the processing of S26 to the processing of S22 and the period from the processing of S22 to the processing of S26.

Next, the CPU 112 acquires time-series data including a set of three sampling values of torque command value Trq *, torque Trq, and acceleration Gx in the selected episode, and time-series data of the state s and the action a ( S44). FIG. 3 shows that the values in parentheses that differ are the values of the variables at different sampling timings. For example, the torque command value Trq * (1) and the torque command value Trq * (2) have different sampling timings. Further, the time-series data of the action a belonging to the selected episode is defined as the action set Aj, and the time-series data of the state s belonging to the same episode is defined as the state set Sj.

Next, the CPU 112 satisfies the condition (a) that the absolute value of the difference between the arbitrary torque Trq belonging to the selected episode and the torque command value Trq * is the specified amount ΔTrq or less, and the acceleration Gx is the lower limit value GxL or more. It is determined whether or not the logical product with the condition (a) indicating that the upper limit value is GxH or less is true (S46).

Here, the CPU 112 variably sets the specified amount ΔTrq according to the amount of change ΔPA per unit time of the accelerator operation amount PA at the start of the episode. That is, when the absolute value of the change amount ΔPA of the accelerator operation amount PA at the start of the episode is large, the CPU 112 considers that the episode is related to the transient time, and sets the specified amount ΔTrq to a larger value than the episode related to the steady state. Set.

Further, the CPU 112 variably sets the lower limit value GxL according to the change amount ΔPA of the accelerator operation amount PA at the start of the episode. That is, when the CPU 112 is an episode related to the transient time and the amount of change ΔPA is positive, the CPU 112 sets the lower limit value GxL to a larger value than the case of the episode related to the steady state. Further, when the CPU 112 is an episode related to the transient time and the amount of change ΔPA is negative, the CPU 112 sets the lower limit value GxL to a smaller value than the case of the episode related to the steady state.

Further, the CPU 112 variably sets the upper limit value GxH according to the change amount ΔPA per unit time of the accelerator operation amount PA at the start of the episode. That is, when the CPU 112 is an episode related to the transient time and the amount of change ΔPA is positive, the CPU 112 sets the upper limit value GxH to a larger value than the case of the episode related to the steady time. Further, when the CPU 112 is an episode related to the transient time and the amount of change ΔPA is negative, the CPU 112 sets the upper limit value GxH to a smaller value than the case of the episode related to the steady state.

When the CPU 112 determines that the logical product is true (S46: YES), it substitutes "10" for the reward r (S48), while when it determines that the logical product is false (S46: NO), the reward r is "". -10 ”is substituted (S50). When the processing of S48 and S50 is completed, the CPU 72 is a group to which the vehicle that transmits the data received by the processing of S40 belongs in order to update the relational regulation data DR for each group that classifies the vehicles VC1, VC2, ... (S52).

FIG. 4 illustrates the classification of vehicles in this embodiment. In the present embodiment, the group is specified by the area where the vehicle is located and the mileage Lt. For example, the vehicle located in the area 1 and the vehicle located in the area 2 belong to different groups from each other. Here, the classification by region may be, for example, region 1 as North America, region 2 as South America, or the like. The classification by region is provided in view of the fact that the optimum action a for the state s may differ because the fuel properties and environment differ from region to region. Further, even if the vehicles belong to the same area, the vehicles having a mileage Lt larger than "0" and "L1" or less and the vehicles having a mileage Lt larger than "L1" and "L2" or less are in different groups. Belongs. In the classification based on the mileage Lt, it is considered that the optimum action a for the state s may differ depending on the degree of wear of the vehicle, for example, the responsiveness of the vehicle to the driving operation of the user may differ. It was provided in consideration of this.

The CPU 72 identifies the area by the position data Pgps, and specifies the group by the specified area and the mileage Lt. Then, the CPU 72 updates the relational regulation data DR common to the specified group. In this embodiment, the ε soft policy on-type Monte Carlo method is used.

That is, the CPU 112 adds the reward r to the profit R (Sj, Aj) determined by the set of each state read by the process of S44 and the corresponding action (S54). Here, "R (Sj, Aj)" is a general description of the profit R in which one of the elements of the state set Sj is a state and one of the elements of the action set Aj is an action. Next, each of the revenues R (Sj, Aj) determined by the set of each state read by the process of S44 and the corresponding action is averaged and substituted into the corresponding action value function Q (Sj, Aj) ( S56). Here, the averaging may be a process of dividing the profit R calculated by the process of S54 by the value obtained by adding a predetermined number to the number of times the process of S54 is performed. The initial value of the profit R may be the initial value of the corresponding action value function Q.

Next, the CPU 112 describes the throttle opening command value TA * and the retard angle amount aop when the maximum values of the corresponding action value functions Q (Sj, A) are obtained for the state read by the process of S44. The set of actions is substituted into the action Aj * (S58). Here, "A" indicates an arbitrary action that can be taken. The action Aj * has a different value depending on the type of the state read by the process of S44, but here, the notation is simplified and described by the same symbol.

Next, the CPU 112 updates the corresponding policy π (Aj | Sj) for each of the states read by the process of S44 (S60). That is, assuming that the total number of actions is "| A |", the selection probability of the action Aj * selected by S58 is "1-ε + ε / | A |". Further, the selection probabilities of "| A | -1" actions other than the action Aj * are set to "ε / | A |", respectively. Since the process of S60 is a process based on the action value function Q updated by the process of S56, the relationship rule data DR that defines the relationship between the state s and the action a increases the profit R. Will be updated to.

Next, the CPU 112 determines whether or not the processing of S44 to S60 has been completed for all the episodes belonging to the data for one trip received by the processing of S40 (S62). Then, when the CPU 112 determines that there is an episode that has not been completed yet (S62: NO), the CPU 112 returns to the process of S42.

On the other hand, when the CPU 112 determines that the processing of S44 to S60 has been completed for all episodes (S62: YES), the CPU 112 operates the communication device 117 to receive the updated relational regulation data DR by the processing of S40. It is transmitted to the data source (S64). When the CPU 112 completes the process of S64, the CPU 112 temporarily ends the series of processes shown in FIG. 3 (b).

On the other hand, as shown in FIG. 3A, the CPU 72 receives the updated relational regulation data DR (S34), and the received relational regulation data DR uses the relational regulation data DR used for the processing of S12. Rewrite (S36). The CPU 72 temporarily ends a series of processes shown in FIG. 3A when the process of S36 is completed or when a negative determination is made in the process of S30.

The relational regulation data DR stored in the storage device 76 at the time of shipment of the vehicle VC1 is subjected to reinforcement learning to some extent by processing according to FIGS. 2 and 3 while simulating the running of the vehicle on a test bench or the like. Let it be a trained model. However, in the pre-shipment learning, the independent variables of the behavioral value function Q, which is the target of reinforcement learning, were partially reduced by human knowledge, etc., out of all the combinations of possible values of the state s and the behavior a. It shall be. That is, for example, in the time series data of the accelerator operation amount PA, one of the two adjacent sampling values becomes the minimum value of the accelerator operation amount PA and the other becomes the maximum value from the operation of the accelerator pedal 86 by a person. The behavioral value function Q is not defined only for pre-shipment learning. In the present embodiment, the possible values of the state s that defines the behavioral value function Q are set to 10 4 or less, more preferably 10 only in the learning before shipment by dimensionality reduction based on human knowledge or the like. Limit to 3 or less.

Here, the operation and effect of this embodiment will be described.
The CPU 72 acquires time-series data of the accelerator operation amount PA as the state s in accordance with the operation of the accelerator pedal 86 by the user, and according to the policy π, the action a consisting of the throttle opening degree command value TA * and the retard angle amount aop. To set. Here, the CPU 72 basically selects the action a that maximizes the expected return based on the action value function Q defined in the relational rule data DR. However, the CPU 72 searches for the action a that maximizes the expected return by selecting an action other than the action a that maximizes the expected return with a predetermined probability ε. Then, the CPU 72 transmits the state s, the action a and the torque Trq, the torque command value Trq *, and the acceleration Gx to the data analysis center 110.

As a result, in the data analysis center 110, the relational regulation data DR can be updated by reinforcement learning as the user drives the vehicle VC1. Therefore, the throttle opening command value TA * and the retard angle amount aop according to the accelerator operation amount PA can be set to appropriate values in the traveling of the vehicle VC1 without excessively increasing the man-hours by the expert. In particular, since the data analysis center 110 updates one relational regulation data DR based on the data transmitted from a plurality of vehicles, the data analysis center 110 updates one relational regulation data DR based on the data transmitted from one vehicle. It is possible to increase the update frequency of the related regulation data DR as compared with. Further, this means that the number of searches for updating the related regulation data DR can be increased in a short time.

That is, the probability that each of the actions other than the greedy action is selected in each state is "ε / | A |", but until the behavior other than the greedy action is sufficiently evaluated in each state, that The action needs to be selected a certain number of times. This means that it is necessary to increase the number of occurrences of each state s. Here, the total frequency of appearance of the specific state s in one million vehicles is about one million times that of the frequency of appearance of the specific state s in one vehicle. Therefore, the time required for the state s to appear as many times as required for the evaluation of the behavior selected by the search to be sufficient is only one vehicle when using the data from one million vehicles. It is about one millionth of the case where the data of is used. Therefore, it is possible for various users to quickly find an appropriate action a when driving a vehicle.

According to the present embodiment described above, the effects described below can be further obtained.
(1) Instead of updating a single relational regulation data DR using all vehicle data at the data analysis center 110, each separate relational regulation data DR was updated for each grouped vehicle data. This makes it possible to update the relational regulation data to more appropriate data for each group even if the relational regulation data DR common to all vehicles is implemented at the time of shipment of the vehicle.

(2) Before shipping, only a part of the value of the independent variable of the action value function Q is learned by reinforcement learning, and the action value of all the values of the independent variable is changed with the operation of the vehicle VC1 after shipping. The function Q can be learned. As a result, based on the enormous amount of data transmitted from a plurality of vehicles VC1, VC2, ..., The behavioral value function Q having an independent variable with a large number of dimensions that is difficult to learn before shipping the product is learned by reinforcement learning. it can.

(3) The time series data of the accelerator operation amount PA was included in the independent variable of the action value function Q. As a result, the value of the action a can be finely adjusted for various changes in the accelerator operation amount PA, as compared with the case where only a single sampling value is used as the independent variable for the accelerator operation amount PA.

(4) The throttle opening command value TA * itself is included in the independent variable of the action value function Q. This makes it easier to increase the degree of freedom of search by reinforcement learning, for example, as compared with the case where the parameters of the model formula that models the behavior of the throttle opening command value TA * are used as independent variables related to the throttle opening. Is.

<Second embodiment>
Hereinafter, the second embodiment will be described with reference to the drawings, focusing on the differences from the first embodiment.

In the present embodiment, the probability ε of taking an action other than the greedy action is gradually reduced with time to zero, so that only the greedy action is selected when a certain amount of time elapses after the vehicle is shipped. To do. Further, in the present embodiment, the classification based on the mileage Lt is performed in only one step, and the re-search is temporarily executed in the group in which the mileage Lt becomes large to some extent.

FIG. 5 shows a processing procedure of reinforcement learning according to the present embodiment. The process shown in FIG. 5A is realized by the CPU 72 executing the learning subprogram 74b stored in the ROM 74 shown in FIG. Further, the process shown in FIG. 5B is realized by the CPU 112 executing the learning main program 114a stored in the ROM 114. In the following, the same step numbers are assigned to the processes corresponding to the processes shown in FIG. 3 for convenience. Hereinafter, the processing shown in FIG. 5 will be described in chronological order.

As shown in FIG. 5B, the CPU 112 executes the processes S40 to S62. Next, the CPU 112 determines whether or not the mileage Lt is equal to or greater than the consumption lower limit value LtL (S70). Here, the consumption lower limit value LtL is a value that determines a threshold value for classification according to the mileage Lt. Further, the lower limit of consumption LtL is set to a value at which the probability ε of selecting an action other than the greedy action is assumed to be zero.

When the CPU 112 determines that the consumption lower limit value LtL or more (S70: YES), the CPU 112 determines whether or not the re-search in the group of vehicles having the consumption lower limit value LtL or more is completed (S72). Here, the completion of the re-search means that the probability ε that an action other than the greedy action by the process of S60a described later is selected is zero. When the CPU 112 determines that the re-search has not been completed yet (S72: NO), the CPU 112 substitutes one action ai other than the greedy action into the search action ae (S74). Next, the CPU 112 updates the label variable i of the action other than the greedy action (S76). Then, the CPU 72 updates the relational regulation data DR for the group whose mileage Lt is equal to or greater than the consumption lower limit value LtL (S60a). Here, the probability that the greedy action is selected is "1-ε", the probability that the search action ae is selected is "ε", and the probability that other actions are selected is "0".

Then, the CPU 112 determines whether or not the related regulation data DR has been updated when the processing of S60a is completed, when a negative determination is made in the processing of S70, or when an affirmative determination is made in the processing of S72 (S78). ). Here, the CPU 112 is a relational provision for a group in which the mileage Lt is equal to or greater than the lower limit of consumption LtL for the vehicle that transmits the data received by the processing of S40 even when the affirmative determination is made by the processing of S72. If the data DR has not been transmitted, it is determined that there is updated data. Then, when the CPU 72 determines that there is updated data (S78: YES), the CPU 72 operates the communication device 117 to transmit the relevant regulation data DR to the vehicle that transmits the data received by the processing of S40. (S64).

The CPU 112 temporarily ends a series of processes shown in FIG. 5B when the process of S64 is completed or when a negative determination is made in the process of S78.
On the other hand, as shown in FIG. 5A, the CPU 72 determines whether or not there is updated data (S80). Then, when the CPU 72 determines that there is updated data (S80: YES), the CPU 72 shifts to the process of S34. The CPU 72 temporarily ends a series of processes shown in FIG. 5A when the process of S36 is completed or when a negative determination is made in the processes of S30 and S80.

Here, the operation and effect of this embodiment will be described.
The CPU 72 acquires time-series data of the accelerator operation amount PA in accordance with the operation of the accelerator pedal 86 by the user, and sets the action a including the throttle opening degree command value TA * and the retard angle amount aop according to the policy π. Here, the CPU 72 basically selects the action a that maximizes the expected return based on the action value function Q defined in the relational rule data DR. However, the CPU 72 searches for the action a that maximizes the expected return by selecting an action other than the action a that maximizes the expected return with a predetermined probability ε. As a result, the related regulation data DR can be updated by reinforcement learning as the user drives the vehicle VC1. As a result, the throttle opening command value TA * and the retard angle amount aop according to the accelerator operation amount PA can be set to appropriate values in the traveling of the vehicle VC1 without excessively increasing the man-hours by the expert.

Here, since the probability ε gradually decreases with the passage of time and becomes zero, when learning is performed to some extent, it is possible to select only the greedy action, assuming that the relational regulation data DR is appropriate for actual driving. .. As a result, the optimum throttle opening command value TA * and the retard angle amount aop are always set.

However, when the mileage Lt of the vehicle becomes large, the related regulation data DR may not be the optimum data due to the consumption of the vehicle. For example, when deposits are deposited on the throttle valve 14 and the intake passage 12, the flow path cross-sectional area of the intake passage 12 becomes small, so that the intake air amount Ga becomes small even if the throttle opening degree TA is the same. Therefore, the throttle opening degree command value TA * that maximizes the expected profit according to the time-series data of the accelerator operation amount PA specified by the related regulation data DR is caused by the accumulation of deposits on the throttle valve 14. It is desirable that the value compensates for the reduction in the cross-sectional area of the flow path of the intake passage 12.

Therefore, in the present embodiment, the re-search is executed for the vehicle in which the mileage Lt is increased to some extent and the vehicle is considered to be exhausted. As a result, after the product is shipped, the mileage Lt of some vehicles becomes the consumption lower limit value LtL or more, so that the relational regulation data DR appropriate for the consumed vehicle can be learned by reinforcement learning.

According to the present embodiment described above, the effects described below can be further obtained.
(4) When performing the re-search process, the CPU 112 changes the search action ae each time the related regulation data DR is transmitted. As a result, as shown in FIG. 6, for example, in the vehicle VC1, the action a1 is adopted as the search action ae, in the vehicle VC2, the action a2 is selected as the search action, and in the vehicle VC3, the action ae is performed. a3 is selected. As a result, it becomes possible to try an action other than the greedy action at a predetermined number of times at an early stage as compared with the case where the vehicle executing the re-search randomly selects the action a.

<Correspondence>
The correspondence between the matters in the above-described embodiment and the matters described in the above-mentioned "means for solving the problem" column is as follows. In the following, the correspondence is shown for each number of the solution means described in the column of "Means for solving the problem". [1, 6, 7] The electronic device corresponds to the throttle valve 14 and the ignition device 26. The first executor corresponds to the CPU 72 and ROM 74, and the second executor corresponds to the CPU 112 and ROM 114. The storage device corresponds to the storage devices 76 and 116. The acquisition process corresponds to the processes of S10 and S16. The operation process corresponds to the process of S14. The reward calculation process corresponds to the processes of S46 to S50. The update process corresponds to the processes of S54 to S60. The updated map corresponds to the map defined by the processing of S54 to S60 defined by the learning main program 114a. [2, 3] The specific process corresponds to the process of S52. [4] The degree of wear is quantified by the mileage Lt. [5] The search process corresponds to the processes of S12 and S14 when the greedy action is not taken. The instruction processing corresponds to the processing of S74, S76, S60a, and S64.

<Other Embodiments>
In addition, this embodiment can be implemented by changing as follows. The present embodiment and the following modified examples can be implemented in combination with each other within a technically consistent range.

・ "Variables indicating the degree of wear"
In the above embodiment, the mileage Lt is exemplified as a variable indicating the degree of wear of the vehicle, but the present invention is not limited to this. For example, it may be the average value of the intake air amount Ga per 1% of the opening degree of the throttle valve 14. Further, for example, as described in the column of "About the vehicle" below, in the case of a vehicle equipped with a rotating electric machine as a thrust generating device of the vehicle, the amount of fully charged charge of the battery that supplies electric power to the rotating electric machine may be used.

・ "About vehicle classification"
In the above embodiment, a plurality of vehicles are divided into a plurality of groups according to the degree of vehicle wear and the area, but the present invention is not limited to this. For example, two factors, the degree of vehicle wear and the area, may be classified according to only one of them.

Further, as a variable for specifying the group, for example, the maximum value or the average value of the accelerator operation amount PA may be used. This makes it possible to classify according to the driving preference of the user.
The classification of vehicles is not limited to those defined in advance from a predetermined viewpoint. For example, a plurality of groups may be automatically generated by unsupervised learning by clustering information obtained from a plurality of vehicles.

By the way, classifying vehicles is not essential.
・ "Instruction processing"
The instruction processing is not limited to the one illustrated by the processing of S74, S76, S60a, and S64. For example, only for a vehicle whose degree of wear is equal to or higher than a predetermined value, an action of increasing the throttle opening command value TA * and an action of decreasing the retard angle amount aop are preferentially executed with respect to the accelerator operation amount PA. It may be a process of instructing. Specifically, for example, after updating the policy π so that the action of making the throttle opening command value TA * smaller than the current greedy action and the action of increasing the retard angle amount aop are not included in the search. The relevant regulation data DR may be transmitted from the data analysis center 110 to the target vehicle.

In addition, for example, when it is discovered that there is an action that significantly increases the expected return among actions other than the current greedy action, it is instructed to execute the found action and an action close to it as a search process. It may be a process to be performed. Here, the action that significantly increases the expected return in a predetermined state may be, for example, an action in which the amount of increase in the action value function Q in a predetermined period is equal to or greater than a predetermined value. Further, in the instruction processing, specifically, the absolute value of the difference between the values of the throttle opening command value TA * is equal to or less than the predetermined value and the absolute value of the difference in the retard angle amount aop is a predetermined value with respect to the discovered action a. The policy π may be updated so that only the following actions are included in the search, and the updated related regulation data DR may be transmitted from the data analysis center 110 to each vehicle.

・ "About behavior variables"
In the above embodiment, the throttle opening command value TA * is exemplified as a variable related to the opening degree of the throttle valve as an action variable, but the present invention is not limited to this. For example, the responsiveness of the throttle opening command value TA * to the accelerator operation amount PA is expressed by the wasted time and the second-order lag filter, and the wasted time and the two variables that define the second-order lag filter are a total of three. The variable may be a variable related to the opening degree of the throttle valve. However, in that case, it is desirable that the state variable is a change amount per unit time of the accelerator operation amount PA instead of the time series data of the accelerator operation amount PA.

In the above embodiment, the retard angle amount aop is exemplified as a variable related to the ignition timing as an action variable, but the present invention is not limited to this. For example, it may be the ignition timing itself to be corrected by KCS.

In the above embodiment, variables related to the opening degree of the throttle valve and variables related to the ignition timing are exemplified as behavioral variables, but the present invention is not limited to this. For example, the fuel injection amount may be used in addition to the variables related to the opening degree of the throttle valve and the variables related to the ignition timing. Further, regarding these three, only the variable related to the opening degree of the throttle valve and the fuel injection amount may be adopted as the action variables, or only the variable related to the ignition timing and the fuel injection amount may be adopted. Furthermore, for those three, only one of them may be adopted as the behavioral variable.

Further, as described in the column of "About the internal combustion engine", in the case of a compression ignition type internal combustion engine, a variable related to the injection amount is used instead of the variable related to the opening degree of the throttle valve, and a variable related to the ignition timing is used instead of the variable related to the ignition timing. You can use the variables related to. In addition to the variables related to the injection timing, the variables related to the number of injections in one combustion cycle and the end timing of one of the two fuel injections adjacent in time series for one cylinder in one combustion cycle and the other It is desirable to add a variable for the time interval between the start timing.

Further, for example, when the transmission 50 is a stepped transmission, the current value of the solenoid valve for adjusting the engaged state of the clutch by flood control may be used as an action variable.
Further, for example, when a hybrid vehicle, an electric vehicle, or a fuel cell vehicle is adopted as the vehicle as described in the column of "About the vehicle" below, the torque and output of the rotary electric machine may be used as an action variable. Further, for example, when an in-vehicle air conditioner including a compressor that is rotated by the rotational power of the crankshaft of the internal combustion engine is provided, the load torque of the compressor may be included in the action variable. Further, when an electric in-vehicle air conditioner is provided, the power consumption of the air conditioner may be included in the action variable.

・ "About the condition"
In the above embodiment, the time-series data of the accelerator operation amount PA is set to the data consisting of six values sampled at equal intervals, but the present invention is not limited to this. The data may be data consisting of two or more sampling values at different sampling timings, and at this time, it is more desirable that the data consists of three or more sampling values or the sampling intervals are evenly spaced.

The state variable related to the accelerator operation amount is not limited to the time series data of the accelerator operation amount PA, but is, for example, the amount of change of the accelerator operation amount PA per unit time as described in the column of "behavior variable". May be good.

Further, for example, as described in the column of "About the action variable", when the current value of the solenoid valve is used as the action variable, the state includes the rotation speed of the input shaft 52 of the transmission, the rotation speed of the output shaft 54, and the solenoid valve. The oil pressure adjusted by may be included. Further, for example, as described in the column of "behavior variable", when the torque or output of the rotary electric machine is used as the action variable, the charge rate or temperature of the battery may be included in the state. Further, for example, as described in the column of "behavior variables", when the load torque of the compressor and the power consumption of the air conditioner are included in the behavior, the temperature inside the vehicle interior may be included in the state.

・ "About dimensionality reduction of table format data"
The method for reducing the dimension of the table format data is not limited to the one illustrated in the above embodiment. For example, since it is rare that the accelerator operation amount PA becomes the maximum value, the action value function Q is not defined for the state where the accelerator operation amount PA becomes the specified amount or more, and the accelerator operation amount PA becomes the specified amount or more. In this case, the throttle opening command value TA * and the like may be adapted separately. Further, for example, the dimension may be reduced by excluding the value at which the throttle opening degree command value TA * is equal to or more than the specified value from the values that can be taken.

It is not essential to expand the range of possible values of the independent variable of the action value function Q, which is the target of reinforcement learning, after the vehicle is shipped, as compared with before the shipment. In addition, it is not essential to carry out reinforcement learning before shipping. For example, after setting the initial value of the action value function Q by diverting the conformity data of a vehicle having the same displacement of an internal combustion engine and having already adapted the behavior variable to the state, the vehicle is shipped and shipped. Reinforcement learning may be performed for the first time later.

・ "Regarding related regulation data"
In the above embodiment, the action value function Q is a table-type function, but the present invention is not limited to this. For example, a function approximator may be used.

For example, instead of using the action value function Q, the policy π may be set as the probability of taking the action a, and the policy π itself may be updated according to the reward r. This can be achieved, for example, by using a function approximator in which the state s and the action a are independent variables and the probability of taking the action a is the dependent variable. In that case, the parameters that determine the function approximator may be updated according to the reward r.

・ "About operation processing"
For example, as described in the column of "Relationship regulation data", when the action value function is used as a function approximation device, all the discrete value sets for the action that are the independent variables of the function of the table type in the above embodiment. The action a that maximizes the action value function Q may be selected by inputting the action value function Q together with the state s.

Further, for example, as described in the column of "Relationship regulation data", when the policy π is a function approximation device in which the state s and the action a are independent variables and the probability of taking the action a is the dependent variable, the policy π is used. Action a may be selected based on the probability indicated by.

・ "About updated mapping"
In the processing of S54 to S60, the method by the ε soft policy on-type Monte Carlo method was illustrated, but the present invention is not limited to this. For example, the policy-off type Monte Carlo method may be used. However, it is not limited to the Monte Carlo method, for example, the policy-off type TD method is used, the policy-on type TD method such as the SARSA method is used, and the eligibility tracing method is used as the policy-on type learning, for example. You may use it.

Further, for example, as described in the column of "Relationship regulation data", when the policy π is expressed by using a function approximation device and this is directly updated based on the reward r, the policy gradient method or the like is used to update the policy π. A map may be constructed.

Further, only one of the action value function Q and the policy π is not limited to the one directly updated by the reward r. For example, the action value function Q and the policy π may be updated, respectively, as in the actor-critic method. Further, in the actor-critic method, the value function V may be updated instead of the action value function Q, for example.

・ "About reward calculation processing"
In the process of FIG. 3, the reward is given according to whether or not the logical product of the condition (a) and the condition (b) is true, but the reward is not limited to this. For example, a process of giving a reward depending on whether or not the condition (a) is satisfied and a process of giving a reward depending on whether or not the condition (b) is satisfied may be executed. Further, for example, with respect to two processes, that is, a process of giving a reward depending on whether or not the condition (a) is satisfied and a process of giving a reward depending on whether or not the condition (b) is satisfied, among them. Only one of the processes may be executed.

Further, for example, instead of giving the same reward uniformly when the condition (a) is satisfied, a process may be performed in which a larger reward is given when the absolute value of the difference between the torque Trq and the torque command value Trq * is smaller than when it is large. .. Further, for example, instead of giving the same reward uniformly when the condition (a) is not satisfied, it is also possible to give a smaller reward when the absolute value of the difference between the torque Trq and the torque command value Trq * is large than when it is small. Good.

Further, for example, instead of giving the same reward uniformly when the condition (a) is satisfied, the process may be such that the magnitude of the reward is variable according to the magnitude of the acceleration Gx. Further, for example, instead of giving the same reward uniformly when the condition (a) is not satisfied, the process may be such that the magnitude of the reward is variable according to the magnitude of the acceleration Gx.

In the above embodiment, the reward r is given according to whether or not the drivability standard is satisfied, but the drivability standard is not limited to the above-mentioned one, and for example, whether noise or vibration intensity satisfies the standard. It may be set according to the case. However, the present invention is not limited to this, for example, whether or not the above acceleration meets the standard, whether or not the followability of the torque Trq meets the standard, whether or not the noise meets the standard, and whether or not the vibration intensity meets the standard. It may be any one or more of the ones.

The reward calculation process is not limited to the one in which the reward r is given depending on whether or not the criteria for drivability are satisfied. For example, if the fuel consumption rate meets the criteria, it may be a process that rewards a larger amount than if it does not meet the criteria. Further, for example, it may be a process of giving a larger reward when the exhaust characteristic satisfies the standard than when it does not meet the standard. In addition, when the drivability standard is met, a larger reward is given than when it is not met, when the fuel consumption rate meets the standard, a larger reward is given than when it is not met, and when the exhaust characteristics meet the standard. It may include two or three of the three processes with the process that rewards more than the case where is not satisfied.

Further, for example, when the current value of the solenoid valve of the transmission 50 is used as the action variable as described in the column of "about the action variable", for example, the following three processes (a) to (c) are performed in the reward calculation process. At least one of these processes may be included.

(A) This is a process in which when the time required for switching the gear ratio by the transmission is within a predetermined time, a larger reward is given than when the predetermined time is exceeded.
(B) This is a process in which when the absolute value of the change speed of the rotation speed of the input shaft 52 of the transmission is equal to or less than the input side predetermined value, a larger reward is given than when the input side predetermined value is exceeded.

(C) This is a process in which when the absolute value of the change speed of the rotation speed of the output shaft 54 of the transmission is equal to or less than the predetermined value on the output side, a larger reward is given than when the predetermined value on the output side is exceeded.
In addition, for example, as described in the column of "About behavior variables", when the torque or output of a rotating electric machine is used as a behavior variable, a process of giving a larger reward than when the battery charge rate is within a predetermined range. Alternatively, it may include a process that rewards a greater amount than if the battery temperature is within a predetermined range. In addition, for example, as described in the column of "About behavior variables", when the load torque of the compressor and the power consumption of the air conditioner are included in the behavior variables, it is larger than when the temperature inside the vehicle is not within the predetermined range. A rewarding process may be added.

・ "About vehicle control system"
The vehicle control system is not limited to the one composed of the control device 70 and the data analysis center 110. For example, it may be composed of a control device 70, a mobile terminal, and a data analysis center 110. This can be realized by the mobile terminal executing the process of S12 or the like.

・ "About the execution device"
The execution device is not limited to the one provided with the CPU 72 (112) and the ROM 74 (114) to execute software processing. For example, a dedicated hardware circuit such as an ASIC that performs hardware processing on at least a part of what has been software-processed in the above embodiment may be provided. That is, the executing device may have any of the following configurations (a) to (c). (A) A processing device that executes all of the above processing according to a program and a program storage device such as a ROM that stores the program are provided. (B) A processing device and a program storage device that execute a part of the above processing according to a program, and a dedicated hardware circuit that executes the remaining processing are provided. (C) A dedicated hardware circuit for executing all of the above processes is provided. Here, there may be a plurality of software execution devices including a processing device and a program storage device, and a plurality of dedicated hardware circuits.

・ "About storage device"
In the above embodiment, the storage device for storing the related regulation data DR and the storage device (ROM74) for storing the learning subprogram 74b and the control program 74a are different storage devices, but the present invention is not limited to this. Further, the storage device for storing the related regulation data DR and the storage device (ROM114) for storing the learning main program 114a are used as separate storage devices, but the present invention is not limited to this.

・ "About internal combustion engine"
The internal combustion engine is not limited to one having a port injection valve for injecting fuel into the intake passage 12 as a fuel injection valve, and may be provided with an in-cylinder injection valve for directly injecting fuel into the combustion chamber 24. For example, it may include both a port injection valve and an in-cylinder injection valve.

The internal combustion engine is not limited to the spark ignition type internal combustion engine, and may be, for example, a compression ignition type internal combustion engine that uses light oil or the like as fuel.
・ "About the vehicle"
The vehicle is not limited to a vehicle in which the thrust generator is only an internal combustion engine, and may be, for example, a so-called hybrid vehicle including an internal combustion engine and a rotary electric machine. Further, for example, as the thrust generator, there may be a so-called electric vehicle or a fuel cell vehicle equipped with a rotating electric machine without providing an internal combustion engine.

10 ... Internal combustion engine, 12 ... Intake passage, 14 ... Throttle valve, 16 ... Fuel injection valve, 18 ... Intake valve, 20 ... Cylinder, 22 ... Piston, 24 ... Combustion chamber, 26 ... Ignition device, 28 ... Crankshaft, 30 ... exhaust valve, 32 ... exhaust passage, 34 ... catalyst, 40 ... torque converter, 42 ... lockup clutch, 50 ... transmission, 52 ... input shaft, 54 ... output shaft, 60 ... drive wheel, 70 ... control device, 72 ... CPU, 74 ... ROM, 74a ... Control program, 74b ... Learning subprogram, 76 ... Storage device, 77 ... Communication device, 78 ... Peripheral circuit, 79 ... Local network, 80 ... Airflow meter, 82 ... Throttle sensor, 84 ... Crank angle sensor, 86 ... Accelerator pedal, 88 ... Accelerator sensor, 90 ... Acceleration sensor, 100 ... Network, 110 ... Data analysis center, 112 ... CPU, 114 ... ROM, 114a ... Main program for learning, 116 ... Storage device, 117 ... communication device, 118 ... peripheral circuit, 119 ... local network.

Claims (7)

A storage device that stores relationship regulation data that defines the relationship between the state of the vehicle and an action variable that is a variable related to the operation of electronic devices in the vehicle, a first execution device mounted on the vehicle, and an in-vehicle device. It is equipped with a second execution device, which is different from the above.
The acquisition process for acquiring the detection value of the sensor that detects the state of the vehicle, and
The operation process for operating the electronic device and
Based on the detected value acquired by the acquisition process, a reward calculation process that gives a larger reward than when the characteristics of the vehicle meet the criteria and does not meet the criteria.
Input of the vehicle state based on the detected value acquired by the acquisition process, the value of the action variable used for the operation of the electronic device, and the reward corresponding to the operation into a predetermined update map. And the update process to update the related regulation data,
Is executed in collaboration with the first executing device and the second executing device.
The first execution device executes at least the acquisition process and the operation process,
The operation process includes a process of operating the electronic device according to the relational regulation data updated by the update process and the state of the vehicle.
The updated map outputs the relevant regulation data updated so as to increase the expected profit for the reward when the electronic device is operated according to the relevant regulation data.
In the update process, the state of the plurality of vehicles, the values of the action variables used for the operation of the electronic devices of the plurality of vehicles, and the reward corresponding to the operation are input to the update map. 2 Vehicle control system executed by the execution device. In the update process, the states of the plurality of vehicles belonging to the same group and the electronic devices of the plurality of vehicles belonging to the same group in order to update the related regulation data for each group in which the vehicles are classified. The process of updating the relational regulation data belonging to the group by inputting the value of the action variable used for the operation and the reward corresponding to the operation into the update map is included.
The vehicle control system according to claim 1, wherein the second execution device executes a specific process for specifying which of the plurality of the groups the state of the vehicle, the value of the action variable, and the reward belong to. The vehicle control system according to claim 2, wherein the specific process includes a process of specifying which group the vehicle belongs to according to the position of the vehicle. The vehicle control system according to claim 2 or 3, wherein the specific process includes a process of specifying which group the vehicle belongs to according to the degree of wear of the vehicle. The operation process includes a search process for executing an operation different from the operation that maximizes the expected return, which is grasped from the relational regulation data.
The second execution device is an instruction process for instructing different values of the first vehicle and the second vehicle, which are two of the plurality of vehicles, as the value of the action variable to be executed as the search process. The vehicle control system according to any one of claims 1 to 4. A vehicle control device including the first execution device according to any one of claims 1 to 5. A vehicle learning device including the second execution device according to any one of claims 1 to 5.

Images