JP6900007B2 - Distribution device - Google Patents

Description

Embodiments of the present invention relate to a distribution device.

In the conventional technology, when controlling the behavior of a vehicle such as an automobile, machine learning represented by a deep neural network is used to autonomously control the power, steering device, and braking device without human instruction or operation. By using the arithmetic module using, the function of action decision that substitutes for human instructions and operations is realized. This arithmetic module has a function to determine the output signal from the input signal according to the internal conversion parameter, and by appropriately determining this internal conversion parameter by machine learning, the sensor and object detection function mounted on the automobile By using the output signal of the above, it is possible to determine the control signals of the power, steering device, and braking device corresponding to the action to be taken by the automobile at that time. (See, for example, Patent Document 1.)

Japanese Unexamined Patent Publication No. 2017-21913

It is expected that the above-mentioned arithmetic module can more appropriately control the behavior of the automobile by learning and updating it. It is desirable that the arithmetic module updated in this way be delivered to the automobile at the timing when learning is deepened and the behavior of the automobile can be controlled more appropriately.
However, in the above-mentioned conventional technology, the distribution timing of the arithmetic module has not been examined. For example, if the arithmetic modules with deep learning are distributed to the automobiles all at once, the amount of communication becomes enormous. Was likely to occur.

An object of the present invention is to provide a distribution device capable of reducing the amount of communication during distribution of an arithmetic module for controlling the behavior of a vehicle.

One embodiment of the present invention is to generate vehicle control information by a plurality of layered arithmetic modules including a neuro arithmetic module that performs arithmetic by a neural network, and a detection result in which a situation around the vehicle is detected. This is a distribution device that distributes the calculation module to an in-vehicle control device that controls the behavior of the vehicle, and the detection result is stored in a storage unit that stores the detection result and the control information acquired from the vehicle. And the information acquisition unit that acquires the control information, the learning unit that learns the neuro calculation module based on the detection result and the control information acquired by the information acquisition unit, and the learning unit that updates the learning unit. Among the updated arithmetic modules that are the neuro arithmetic modules, the selection unit that selects the updated arithmetic module that satisfies a predetermined selection condition as the distribution target arithmetic module, and the distribution target arithmetic module selected by the selection unit. Is a distribution device including a distribution unit that distributes the above-mentioned vehicle-mounted control device.

Further, in the distribution device according to the embodiment of the present invention, the predetermined selection condition is a condition based on the number of times of learning for the calculation module, and the selection unit has a threshold value for the number of times of learning for the calculation module. The updated arithmetic module exceeding the above is selected as the distribution target arithmetic module.

Further, in the distribution device according to the embodiment of the present invention, the predetermined selection condition includes an output value output by the pre-update calculation module, which is the pre-update calculation module before the update by learning of the learning unit, and the post-update calculation. It is a condition by comparison with the output value output by the module, and in the selection unit, the difference between the output value output by the pre-update calculation module and the output value output by the post-update calculation module is a predetermined determination criterion. The post-update calculation module that exceeds the number is selected as the distribution target calculation module.

Further, in the distribution device according to the embodiment of the present invention, the selection unit uses the output values output by the plurality of layered arithmetic modules including the pre-update arithmetic module and the pre-update arithmetic module after the update. The distribution target arithmetic module is selected based on the difference from the output values output by the plurality of arithmetic modules replaced by the arithmetic modules.

Further, in the distribution device according to the embodiment of the present invention, when the updated calculation module exists in a plurality of layers, the selection unit selects the distribution target calculation module for each layer.

Further, in the distribution device according to the embodiment of the present invention, the selection unit exists in the upper layer of the hierarchy after selecting the updated arithmetic module existing in the lower layer of the hierarchy as the distribution target arithmetic module. The updated arithmetic module to be distributed is selected as the distribution target arithmetic module.

Further, in the distribution device according to the embodiment of the present invention, the distribution target calculation module is associated with the type of the distribution target vehicle, and the distribution unit is based on the distribution priority according to the vehicle type. The distribution target calculation module selected from the plurality of distribution target calculation modules according to the type of the vehicle is distributed.

According to the present invention, it is possible to provide a distribution device capable of reducing the amount of communication at the time of distribution of an arithmetic module for controlling the behavior of a vehicle.

It is a figure which shows an example of the functional structure of the arithmetic module distribution system of this embodiment. It is a figure which shows an example of the control information of this embodiment. It is a figure which shows an example of the calculation path of the vehicle-mounted control device of this embodiment. It is a figure which shows an example of the processing flow of the vehicle-mounted control device of this embodiment. It is a figure which shows the 2nd example of the process flow of the vehicle-mounted control device of this embodiment. It is a figure which shows the 3rd example of the process flow of the vehicle-mounted control device of this embodiment. It is a figure which shows the specific example of the structure of the arithmetic module of this embodiment. It is a figure which shows an example of the learning procedure of the arithmetic module of this embodiment. It is a figure which shows an example of the selection procedure of the distribution target calculation module by the selection part of this embodiment schematically. It is a figure which shows an example of the operation of the distribution apparatus of this embodiment. It is a figure which shows an example of the learning number of times of the arithmetic module by the learning part of this embodiment. It is a figure which shows an example of the selection operation of the arithmetic module of the selection part of this embodiment. It is a figure which shows an example of the comparison between the arithmetic modules by the selection part of this embodiment. It is a figure which shows an example of the comparison between the arithmetic module group by the selection part in the modification of this embodiment. It is a figure which shows an example of the first-stage comparison between a group of arithmetic modules by a selection part in the modification of this embodiment. It is a figure which shows an example of the comparison of the 2nd stage between the arithmetic module group by the selection part in the modification of this embodiment.

[First Embodiment]
Hereinafter, an outline of the arithmetic module distribution system 1 of the present embodiment will be described with reference to the drawings.
FIG. 1 is a diagram showing an example of a functional configuration of the arithmetic module distribution system 1 of the present embodiment. The calculation module distribution system 1 includes a distribution device 70 that distributes a calculation module to the vehicle CR, a log data storage unit 80, and a module library 90. First, the functional configuration of the vehicle CR will be described.

[Vehicle CR functional configuration]
The vehicle CR includes an in-vehicle control device 10, a detection information generation unit 20, an in-vehicle device 30, a logger 40, a communication unit 50, and a module update unit 60. The in-vehicle control device 10 and the detection information generation unit 20 and the in-vehicle control device 10 and the in-vehicle device 30 are connected to each other by an in-vehicle communication network such as CAN (Control Area Network), and information can be obtained. It is possible to give and receive.
In the present embodiment, the vehicle-mounted control device 10 will be described as a device having a housing independent of the detection information generation unit 20 and the vehicle-mounted device 30, but the present invention is not limited to this. The in-vehicle control device 10 may be configured as a device integrated with the detection information generation unit 20 and the in-vehicle device 30.

The detection information generation unit 20 detects the operation of each part of the vehicle CR and the situation around the vehicle CR, and generates the detection information. The detection information generated by the detection information generation unit 20 is also referred to as a detection result DR. The detection result DR may include the detection result of the operation of each part of the vehicle CR and the detection result of the situation around the vehicle CR, but the detection result of the situation around the vehicle CR will be described below. , The description of the detection result of the operation of each part of the vehicle CR will be omitted.

The detection information generation unit 20 includes a sensor unit 200 and a detection result processing unit 210.
The sensor unit 200 is a sensor or the like that detects the situation around the vehicle CR, and outputs the detected result to the in-vehicle control device 10 as a detection result DR via the detection result processing unit 210 or directly. As an example of this embodiment, the sensor unit 200 includes a lidar device 201, a millimeter wave device 202, a GNSS device 203, and a camera device 204.
The LIDAR device 201 acquires an image of the surroundings of the vehicle CR and measures the distance by LIDAR (Light Detection and Ranging, Laser Imaging Detection and Ranging). The lidar device 201 outputs the image acquisition result and the distance measurement result as the detection result DR.
The millimeter wave device 202 uses millimeter waves to detect the shape around the vehicle CR and measure the distance. The millimeter wave device 202 outputs the shape detection result and the distance measurement result as the detection result DR.
The GNSS device 203 receives a positioning signal from a positioning satellite by a GNSS (Global Navigation Satellite System). The GNSS device 203 outputs the received positioning signal as a detection result DR.
The camera device 204 acquires an image of the surroundings of the vehicle CR and measures the distance using visible light or infrared light. The camera device 204 outputs the image acquisition result and the distance measurement result as the detection result DR.

The detection result processing unit 210 processes the detection result DR output by the sensor unit 200 to generate information indicating the situation around the vehicle CR. In an example of this embodiment, the detection result processing unit 210 includes an image recognition unit 211, an object detection unit 212, and a position estimation unit 213.
Based on the image acquisition result and the distance measurement result output by the lidar device 201 and the camera device 204, the image recognition unit 211 determines the presence / absence of an object included in these images, the shape / size / type of the object, and the distance to the object. , Recognize the moving speed of an object, etc. The image recognition unit 211 outputs the result of image recognition as a detection result DR to the in-vehicle control device 10.
The object detection unit 212 detects the presence / absence of an object, the shape / size / type of the object, the distance to the object, the moving speed of the object, and the like based on the shape detection result and the distance measurement result output by the millimeter wave device 202. .. The object detection unit 212 outputs the result of the object detection as the detection result DR to the in-vehicle control device 10.
The position estimation unit 213 estimates the position of the own vehicle based on the positioning signal output by the GNSS device 203. The position estimation unit 213 outputs the result of the position estimation of the own vehicle as the detection result DR to the in-vehicle control device 10.

The in-vehicle device 30 includes an ECU (Electronic Control Unit) 300 and actuators (not shown), and controls the behavior of the vehicle CR. The ECU 300 includes, for example, a travel control ECU 301 that drives an engine or a travel motor, a brake control ECU 302 that controls a brake, a steering ECU 303 that controls a steering angle, and a light that controls lights such as headlights and emergency stop lights. There is a control ECU 304 and the like.
These ECUs 300 operate based on the control information CMD output from the in-vehicle control device 10, and output information and signals for driving the actuators.

The in-vehicle control device 10 includes a CPU (Central Processing Unit) (not shown), performs an calculation based on the detection result DR generated by the detection information generation unit 20, and generates a control information CMD as a result of this calculation. .. The vehicle-mounted control device 10 of the present embodiment performs this calculation by a calculation module M layered in a plurality of layers.

That is, the in-vehicle control device 10 provides the control information CMD that is given to the in-vehicle device 30 that controls the behavior of the vehicle CR based on the detection result (detection information) DR including at least the information that the surrounding situation of the vehicle CR is detected. It is generated by a plurality of layered arithmetic modules M.
The in-vehicle control device 10 includes an acquisition unit 120 (not shown), an arithmetic module M, and an output unit 130 (not shown) as functional units by software or hardware.

The acquisition unit 120 acquires the detection result DR output by the detection information generation unit 20. The acquisition unit 120 normalizes the acquired detection result DR and supplies the normalized information to the arithmetic module M. For example, when the lidar device 201 is a device that detects 16 directions, the acquisition unit 120 sets the sensor value of the first direction, the sensor value of the second direction, ..., And the sensor value of the 16th direction to 0.0 to 1, respectively. Normalize to values in the range of 0.0. The acquisition unit 120 supplies the normalized values to the calculation module M as a data string in which the normalized values are arranged in the order of the first direction, the second direction, ..., The 16th direction.

The output unit 130 outputs the control information CMD generated by the arithmetic module M to the in-vehicle device 30. An example of the control information CMD output by the output unit 130 will be described with reference to FIG.

FIG. 2 is a diagram showing an example of the control information CMD of the present embodiment. The operations that are possible for the vehicle CR are predefined as control information CMD. The output unit 130 outputs a value "1" when operating each of these control information CMDs, and outputs a value "0 (zero)" when not operating. In the example shown in the figure, the output unit 130 outputs a value "1" for each of the control information CMD "turn the steering wheel (steering wheel) to the right" and the control information CMD "step on the accelerator". As a result of outputting this control information CMD, the ECU 300 of the in-vehicle device 30 controls the steering angle and the power for traveling, so that the vehicle CR "accelerates while turning to the right".

Returning to FIG. 1, the calculation module M includes a selection module 100 and a control module 110.
The selection module 100 selects the calculation module M that performs the calculation according to the situation around the vehicle CR from the plurality of calculation modules M in the next layer based on the detection result DR acquired by the acquisition unit 120. The selection module 100 performs calculations by a neural network NN (for example, a deep neural network). In the following description, the calculation module M that performs calculations by the neural network NN is also referred to as a neuro calculation module MN.

The control module 110 is selected by the selection module 100 and generates control information CMD based on the detection result DR.
That is, the arithmetic module M is layered with the selection module 100 as the arithmetic module M in the previous stage and the control module 110 as the arithmetic module M in the latter stage.

[Control module hierarchy]
The layering of the control module 110 will be described. The layered control module 110 includes a front-stage control module 110a and a rear-stage control module 110b. Of these, the front-stage control module 110a determines the target behavior TB of the vehicle CR based on the detection result DR. The rear-stage control module 110b generates control information CMD to be given to the in-vehicle device 30 based on the target behavior TB of the vehicle CR determined by the front-stage control module 110a.

Here, the target behavior TB of the vehicle CR is "acceleration while turning to the right" in the example shown in FIG. In the case of this example, the vehicle CR is waiting for a right turn in the right turn lane of the intersection, and turns right at the intersection following the progress of the vehicle in front, which is also waiting for the right turn. The detection information generation unit 20 transmits the detection result DR indicating that the vehicle CR is in the right turn lane of the intersection and the detection result DR indicating that the vehicle in front of the vehicle CR has advanced and started a right turn to the in-vehicle control device 10. Supply. Based on these detection result DRs, the front-stage control module 110a determines that the target behavior TB of the vehicle CR is "accelerating while turning to the right". The rear-stage control module 110b "turns the steering wheel (steering wheel) to the right: value 1" and control information based on the target behavior TB of the vehicle CR determined by the front-stage control module 110a, that is, "acceleration while turning to the right". The CMD "step on the accelerator: value 1" is generated as the control information CMD.
In the example shown in FIG. 1, among the control modules 110, the control module 111 and the control module 112 correspond to the front stage control module 110a, and the control module 115 and the control module 116 correspond to the rear stage control module 110b. That is, in the example of the figure, the control module 111 and the control module 115 are layered with each other, and the control module 112 and the control module 116 are layered with each other.

[Layered selection module]
The layering of the selection module 100 will be described. The layered selection module 100 includes a front-stage selection module 100a and a rear-stage selection module 100b. Among them, the front-stage selection module 100a selects the selection module 100 according to the situation around the vehicle CR from the plurality of selection modules 100 in the next layer. The rear-stage selection module 100b is selected by the front-stage selection module 100a, and selects the calculation module M according to the surrounding conditions of the vehicle CR from the plurality of calculation modules M in the next layer.

In the example shown in FIG. 1, in the relationship between the selection module 101 and the selection module 102 among the selection modules 100, the selection module 101 corresponds to the front-stage selection module 100a, and the selection module 102 corresponds to the rear-stage selection module 100b. .. Further, in the relationship between the selection module 102 and the selection module 103, the selection module 102 corresponds to the front-stage selection module 100a, and the selection module 103 corresponds to the rear-stage selection module 100b.

The logger 40 is connected to the vehicle-mounted control device 10, the detection information generation unit 20, and the vehicle-mounted device 30, and temporarily stores the information output by each of these units and each device.

The communication unit 50 transmits / receives information to / from the distribution device 70 via the base station BS. Specifically, the communication unit 50 transmits the information temporarily stored in the logger 40 to the distribution device 70. Further, the communication unit 50 receives the update data of the arithmetic module M provided by the distribution device 70. In the following description, the update data of the arithmetic module M provided by the distribution device 70 will also be described as the distribution target arithmetic module MDT.

When the communication unit 50 receives the distribution target calculation module MDT, the module update unit 60 updates the calculation module M of the vehicle-mounted control device 10 to the distribution target calculation module MDT. In this example, the arithmetic module M of the vehicle-mounted control device 10 is stored in a rewritable non-volatile memory such as a flash memory. In this case, the module update unit 60 rewrites the arithmetic module M already stored in the non-volatile memory to the distribution target arithmetic module MDT received by the communication unit 50.

[Functional configuration of distribution device 70]
The distribution device 70 updates the calculation module M by re-learning the calculation module M based on the log data provided from the vehicle CR, and distributes the updated calculation module M to the vehicle CR. The log data storage unit 80 and the module library 90 are connected to the distribution device 70. The log data storage unit 80 and the module library 90 will be described as being configured as a device separate from the distribution device 70, but the present invention is not limited to this. The distribution device 70, the log data storage unit 80, and the module library 90 may be partially or wholly configured as an integrated device.

The log data storage unit 80 stores the detection result DR and the control information CMD acquired from the vehicle CR. In the following description, the detection result DR and the control information CMD acquired from the vehicle CR are also collectively referred to as log data.
The calculation module M distributed by the distribution device 70 is stored in the module library 90. The distribution device 70 selects an arithmetic module M to be distributed from the arithmetic modules M stored in the module library 90, and distributes the selected arithmetic module M to the vehicle CR.

The distribution device 70 includes a distribution device communication unit 710, an information acquisition unit 720, a learning unit 730, and a selection unit 740.
The distribution device communication unit 710 receives the log data transmitted by the communication unit 50 of the vehicle CR. Further, the distribution device communication unit 710 transmits the distribution target calculation module MDT to the vehicle CR.
The information acquisition unit 720 acquires the detection result DR and the control information CMD from the log data storage unit 80 (storage unit).
The learning unit 730 learns the neuro calculation module MN based on the detection result DR and the control information CMD acquired by the information acquisition unit 720.
The selection unit 740 selects the updated calculation module MAT that satisfies a predetermined selection condition as the distribution target calculation module MDT among the updated calculation module MATs that are the neuro-calculation module MN updated by the learning performed by the learning unit 730. ..

The selection unit 740 includes a distribution target module selection unit 741 and a distribution destination vehicle selection unit 742. The distribution target module selection unit 741 selects the distribution target calculation module MDT from the calculation modules M stored in the module library 90. The delivery destination vehicle selection unit 742 selects the delivery destination vehicle CR of the selected delivery target calculation module MDT from among a plurality of vehicle CRs.

The selection unit 740 supplies the selected distribution target calculation module MDT to the distribution device communication unit 710. The distribution device communication unit 710 distributes the distribution target calculation module MDT selected by the selection unit 740 to the in-vehicle control device 10. When the distribution device communication unit 710 transmits the distribution target calculation module MDT, the distribution device communication unit 710 functions as a distribution unit.

[Calculation path of in-vehicle control device 10]
Next, the calculation path of the in-vehicle control device 10 will be described with reference to FIG.
FIG. 3 is a diagram showing an example of the calculation path of the vehicle-mounted control device 10 of the present embodiment. The selection module 101 acquires the detection result DR from the acquisition unit 120 (not shown). The selection module 101 determines whether to select the selection module 102 or the control module 112 based on the detection result DR. In this example, the selection module 101 selects the selection module 102. Next, the selection module 102 determines whether to select the selection module 103 or the control module 111 based on the detection result DR. In this example, the selection module 102 selects the control module 111. The control module 111 determines the target behavior TB of the vehicle CR and supplies the target behavior TB to the control module 115. The control module 115 generates control information CMD based on the target behavior TB. The generated control information CMD is supplied to each ECU 300 by an output unit 130 (not shown).

[Processing Flow of In-Vehicle Control Device (Example 1)]
FIG. 4 is a diagram showing an example of a processing flow of the in-vehicle control device 10 of the present embodiment. In this example, the vehicle-mounted control device 10 includes one selection module 100 and two control modules 110 as arithmetic modules M. In the case of this example, the arithmetic module M of the vehicle-mounted control device 10 is layered into two layers, a first-stage selection module 100 and a second-stage control module 110.

(Step S10) The sensor unit 200 detects the situation around the vehicle CR and outputs the detected result as the detection result DR.
(Step S20) The detection result processing unit 210 detects the presence / absence of an object, the shape / size / type of the object, the distance to the object, the moving speed of the object, and the like based on the detection result DR output in step S10. .. The detection result processing unit 210 outputs the detected result as the detection result DR to the in-vehicle control device 10.
(Step S30) The acquisition unit 120 of the vehicle-mounted control device 10 acquires the detection result DR output from the sensor unit 200 or the detection result DR output from the detection result processing unit 210.

(Step S40) The selection module 100 selects the control module 110 from the plurality of control modules 110 according to the situation around the vehicle CR indicated by the detection result DR acquired in step S30.
(Step S50) The control module 110 generates control information CMD based on the detection result DR. Here, when the selection module 100 selects the control module 110-A (step S40; A), the control module 110-A executes step S50-1. When the selection module 100 selects the control module 110-B (step S40; B), the control module 110-B executes step S50-2.

(Step S60) Each ECU 300 controls the behavior of the vehicle CR by driving the actuators based on the control information CMD generated in step S50. For example, the travel control ECU 301 drives the engine and the travel motor based on the control information CMD (step S60-1). The brake control ECU 302 controls the brake based on the control information CMD (step S60-2). The steering ECU 303 controls the steering angle based on the control information CMD (step S60-3).

[Processing flow of in-vehicle control device (Example 2)]
Next, a second example of the processing flow of the in-vehicle control device 10 will be described with reference to FIG.
FIG. 5 is a diagram showing a second example of the processing flow of the vehicle-mounted control device 10 of the present embodiment. In the following description, the same parts and the same operations as those described in the example of FIG. 4 are designated by the same reference numerals and the description thereof will be omitted. This embodiment differs from the first embodiment shown in FIG. 4 in that the control module 110 is divided into a plurality of layers.

(Step S150) The front-stage control module 110a generates the target behavior TB of the vehicle CR based on the detection result DR. Here, when the selection module 100 selects the pre-stage control module 110a-A (step S40; A), the pre-stage control module 110a-A executes step S150-1. When the selection module 100 selects the pre-stage control module 110a-B (step S40; B), the pre-stage control module 110a-B executes step S150-2.

(Step S160) The latter stage control module 110b generates control information CMD based on the target behavior TB of the vehicle CR and the detection result DR. Here, when the selection module 100 selects the post-stage control module 110b-A (step S150; A), the post-stage control module 110b-A executes step S160-1. When the selection module 100 selects the post-stage control module 110b-B (step S150; B), the post-stage control module 110b-B executes step S160-2.

[Processing flow of in-vehicle control device (Example 3)]
Next, a third example of the processing flow of the in-vehicle control device 10 will be described with reference to FIG.
FIG. 6 is a diagram showing a third example of the processing flow of the vehicle-mounted control device 10 of the present embodiment. In the following description, the same parts and operations as described in the examples of FIGS. 4 and 5 are designated by the same reference numerals and the description thereof will be omitted. This embodiment differs from the first embodiment shown in FIG. 4 in that the selection module 100 is divided into a plurality of layers.

(Step S240) The front-stage selection module 100a is set to the situation around the vehicle CR indicated by the detection result DR acquired in step S30 from among the plurality of rear-stage selection modules 100b (rear-stage selection modules 100b-A and 100b-B). The corresponding latter-stage selection module 100b is selected.
(Step S250) The latter stage selection module 100b is a vehicle indicated by the detection result DR acquired in step S30 from among a plurality of control modules 110 (control modules 110-A1, 110-A2, 110-B1 and 110-B2). The control module 110 is selected according to the situation around the CR.

The calculation module M may be configured by combining the above-described second and third embodiments.

[Specific example of arithmetic module configuration of in-vehicle control device]
Next, a specific example of the configuration of the calculation module M of the present embodiment will be described with reference to FIG. 7.
FIG. 7 is a diagram showing a specific example of the configuration of the calculation module M of the present embodiment. The selection module 100 is composed of four layers of selection modules 101 to 104. The selection modules 101 to 104 all perform calculations by the neural network NN.

The control module 110 is classified into three control modes MD. The control mode MD includes a Yuzuriai mode MD1, a normal driving mode MD2, and an emergency action mode MD3.
The Yuzuriai mode MD1 is composed of roundabout acceleration / deceleration (control module 111) and lane travel (control module 115).
The normal traveling mode MD2 includes a roundabout / intersection peripheral traveling mode MD2-1 and a straight road traveling mode MD2-2. The roundabout / intersection area traveling mode MD2-1 further includes an acceleration / deceleration mode MD2-1-1 and a stop mode MD2-1-2.
The acceleration / deceleration mode MD2-1-1 is composed of acceleration / deceleration (control module 113a) according to an oncoming vehicle / merging destination vehicle and lane travel (control module 117a).
The stop mode MD2-1-2 is composed of a stop at a stop line (control module 113b) and a lane run (control module 117b).
The straight road traveling mode MD2-2 is composed of acceleration / deceleration (control module 113) according to the preceding vehicle and lane traveling (control module 117).
The emergency action mode MD3 is composed of an emergency stop (control module 112) and lane travel (control module 116).

Here, acceleration / deceleration (control module 111) and emergency stop (control module 112) at the roundabout are both calculated by the neural network NN.
The lane travel (control module 115, control module 116, control module 117, control module 117a, and control module 117b) all perform calculations by a neural network NN and a PID (Proportional-Integral-Differential) control algorithm.
Acceleration / deceleration according to the preceding vehicle (control module 113), acceleration / deceleration according to the oncoming vehicle / merging destination vehicle (control module 113a), and stop at the stop line (control module 113b) are all performed by rule-based calculation. Perform the calculation. The rule-based calculation refers to a calculation procedure for calculating the control information CMD given to the in-vehicle device 30 in which the correspondence relationship between the detection result DR and the control information CMD is predetermined. As an example of the rule-based operation, there is an operation method for calculating the control information CMD by conditional branching by if-then-else.

The selection module 101 selects either the selection module 102 or the control module 110 of the emergency action mode MD3 based on the detection result DR acquired by the acquisition unit 120. As an example, when an emergency situation such as a person jumping out from a side road or a load of a vehicle in front falling occurs while the vehicle CR is running, the selection module 101 selects the control module 110 of the emergency action mode MD3. To do. The selection module 101 selects the selection module 102 if an emergency state such as this example has not occurred.

The selection module 102 selects either the Yuzuriai mode MD1 or the selection module 103 based on the detection result DR. As an example, in a state where the vehicle CR is traveling at a roundabout (roundabout), there is a case where the entire vehicle group can travel efficiently by traveling while the vehicles are swaying each other. In such a case, the selection module 102 selects the Yuzuriai mode MD1. The selection module 102 selects the selection module 103 if it is not traveling at the roundabout.

The selection module 103 selects either the selection module 104 or the straight road traveling mode MD2-2. As an example, when the vehicle CR travels at a roundabout or an intersection, the selection module 103 selects the selection module 104. When the vehicle CR travels on a straight road, the selection module 103 selects the straight road travel mode MD2-2.

The selection module 104 selects either the acceleration / deceleration mode MD2-1-1 or the stop mode MD2-1-2. As an example, when the vehicle CR shows a stop (red) signal at an intersection with a traffic light or travels at an intersection with a stop sign, the selection module 104 sets the stop mode MD2-1-2. select. When the vehicle CR merges in the roundabout, or when the vehicle follows the preceding vehicle in the roundabout, the selection module 104 selects the acceleration / deceleration mode MD2-1-1.

When the control module 110 belonging to each control mode MD is selected by the selection module 100, the calculation result is output to each ECU 300 as control information CMD via the output unit 130.

[Learning procedure of calculation module M]
Next, an example of the learning procedure of the calculation module M performed by the learning unit 730 will be described with reference to FIG.
FIG. 8 is a diagram showing an example of a learning procedure of the calculation module M of the present embodiment. As an example, the control module 110 is composed of two layers, a front-stage control module 110a and a rear-stage control module 110b, and the selection module 100 selects a specific front-stage control module 110a from a plurality of front-stage control modules 110a. The case where it is configured in is described.

In the first stage, learning (for example, machine learning) of the subsequent stage control module 110b is performed (FIG. 8 (A)). The subsequent stage control module 110b includes a PID control algorithm. In the first stage, what kind of value should be set for this PID control parameter is learned.
In the second stage, the learned rear-stage control module 110b is connected to the front-stage control module 110a, and the front-stage control module 110a is learned (FIG. 8 (B)). In this second stage, the system designer determines what kind of driving scene the front-stage control module 110a is for for each front-stage control module 110a, and causes learning based on the data of the driving scene. That is, the pre-stage control module 110a is trained by focusing on a specific driving scene.
In the third stage, the trained rear-stage control module 110b and the front-stage control module 110a are connected to the selection module 100 to train the selection module 100 (FIG. 8C). In this third stage, while observing the behavior of the pre-stage control module 110a with respect to the detection result DR, it is explored which pre-stage control module 110a should be selected for a certain detection result DR. As a result, in the third stage, the selection module 100 is made to learn what kind of driving scene the certain detection result DR shows.

That is, in this example, there are two or more arithmetic modules M in the subsequent stage of a certain arithmetic module M, and after machine learning so that the control information CMD of the power, steering device, and braking device can be determined specifically for different situations, It is connected to the arithmetic module M in the previous stage. The module in the first stage determines the situation from the detection result DR of the sensor unit 200 and the detection result processing unit 210 obtained while the vehicle is running, and selects one of the calculation modules M in the second stage according to the situation at that time. To operate. The operation module M in the subsequent stage determines the control information CMD of the power, the steering device, and the braking device from the detection result DR of the sensor unit 200 and the detection result processing unit 210.

[Selection of calculation module MDT to be distributed]
Next, the procedure for selecting the distribution target calculation module MDT by the selection unit 740 of the distribution device 70 will be described.

FIG. 9 is a diagram schematically showing an example of a procedure for selecting the distribution target calculation module MDT by the selection unit 740 of the present embodiment. The pre-update arithmetic module MBF and the post-update arithmetic module MAT are stored in the module library 90. Here, the pre-update calculation module MBF is the pre-learning calculation module M by the learning unit 730. As an example, the pre-update calculation module MBF is the same calculation module M as the distribution target calculation module MDT that has already been delivered to the vehicle CR. The pre-update calculation module MBF includes a calculation module M corresponding to each calculation module M included in the vehicle-mounted control device 10 of the vehicle CR. For example, the pre-update arithmetic module MBF includes a pre-update arithmetic module MBF1 as a pre-update arithmetic module MBF corresponding to the selection module 101 shown in FIG. 1 and a pre-update arithmetic module corresponding to the selection module 102 shown in FIG. The pre-update calculation module MBF2 as an MBF is included. In this example, the module library 90 stores m types (m is a natural number) of pre-update calculation modules MBF. That is, the module library 90 stores the pre-update calculation module MBF from the pre-update calculation module MBF1 to the pre-update calculation module MBFm. Among these pre-update calculation module MBFs, the k-type (k is a natural number) pre-update calculation module MBF is described as the pre-update calculation module MBFk.

The learning unit 730 (not shown in the figure) learns each of these pre-update calculation module MBFs based on the log data stored in the log data storage unit 80. The learning unit 730 stores the learned calculation module M as a calculation module MAT after updating in the module library 90. Specifically, the learning unit 730 learns the pre-update calculation module MBF1 and stores the learned calculation module M as the post-update calculation module MAT1 in the module library 90. Similarly, the learning unit 730 learns each k (however, 1 ≦ k ≦ m.) For the pre-update calculation module MBFk, and sets the learned calculation module M as the post-update calculation module MATk in the module library. Store in 90.

The learning unit 730 may perform learning on all types of pre-update calculation module MBFs among m types of pre-update calculation module MBFs. Further, there may be a pre-update calculation module MBF in which the learning unit 730 does not learn.
Further, the learning unit 730 may repeatedly (multiple times) learn the calculation module MAT after the update.

The selection unit 740 selects the distribution target calculation module MDT from the updated calculation module MAT stored in the module library 90. In the example of the figure, the selection unit 740 selects the updated calculation module MATk as the distribution target calculation module MDT. The distribution device communication unit 710 distributes (transmits) the updated arithmetic module MATk selected by the selection unit 740 to the vehicle CR as the distribution target arithmetic module MDT.

[Operation of distribution device 70]
An example of the operation of the distribution device 70 of the present embodiment will be described with reference to FIG.
FIG. 10 is a diagram showing an example of the operation of the distribution device 70 of the present embodiment. In the figure, it is assumed that the log data required for learning is stored in the log data storage unit 80, and the calculation module M to be learned is stored in the module library 90.

(Step S110) The learning unit 730 selects the calculation module M to be learned from the calculation modules M stored in the module library 90.
(Step S120) The learning unit 730 acquires log data from the log data storage unit 80.
(Step S130) The learning unit 730 performs learning on the calculation module M (for example, the pre-update calculation module MBF) selected in step S110 based on the log data acquired in step S120. The learning unit 730 stores the learned calculation module M as the updated calculation module MAT in the module library 90.

(Step S140) The selection unit 740 selects the calculation module M to be distributed as the distribution target calculation module MDT among the updated calculation module MATs stored in the module library 90.
(Step S150) The distribution device communication unit 710 distributes (transmits) the distribution target calculation module MDT selected in step S140 to the vehicle CR.
Next, the details of step S140, that is, the selection conditions of the distribution target calculation module MDT by the selection unit 740 will be described.

[Selection conditions for distribution target calculation module MDT (1) Selection based on the number of learnings]
The selection unit 740 selects the distribution target calculation module MDT based on the number of times the calculation module M is learned by the learning unit 730. Hereinafter, a more specific description will be given with reference to FIG.

FIG. 11 is a diagram showing an example of the number of times the calculation module M is learned by the learning unit 730 of the present embodiment. The learning unit 730 records the number of times the calculation module M is learned each time the calculation module M is learned. In the example of the figure, the updated calculation module MAT1 has been learned 500 times. Similarly, learning has been performed 100 times for the updated arithmetic module MAT2, 300 times for the updated arithmetic module MAT3, ..., 1000 times for the updated arithmetic module MATk, ..., 300 times for the updated arithmetic module MATm. ..

The selection unit 740 selects, among the updated arithmetic modules MAT, the arithmetic module M whose learning count exceeds a predetermined threshold value as the distribution target arithmetic module MDT. As an example, when the predetermined threshold value is 800 times, the selection unit 740 selects the updated calculation module MAT that has been learned more than 800 times as the distribution target calculation module MDT. In an example of the figure, the selection unit 740 selects the updated calculation module MATk as the distribution target calculation module MDT as the calculation module M having a learning count exceeding 800 times, which is a predetermined threshold value.

That is, in this example, the predetermined selection condition is a condition based on the number of times of learning for the calculation module M. Further, the selection unit 740 selects the updated arithmetic module MAT whose learning count for the arithmetic module M exceeds a predetermined threshold value as the distribution target arithmetic module MDT.
Here, the selection unit 740 has been described as selecting one post-update calculation module MAT, but the present invention is not limited to this. When there are a plurality of updated arithmetic module MATs in which the number of learnings for the arithmetic module M exceeds a predetermined threshold value, the selection unit 740 selects each of the plurality of updated arithmetic module MATs as the distribution target arithmetic module MDT. May be good.

[Selection conditions for calculation module to be distributed (2) Selection by comparing output values]
The selection unit 740 selects the distribution target calculation module MDT based on the comparison result between the output value of the pre-update calculation module MBF and the output value of the post-update calculation module MAT. Hereinafter, a more specific description will be given with reference to FIG.

FIG. 12 is a diagram showing an example of a selection operation of the calculation module M of the selection unit 740 of the present embodiment.
(Step S210) The selection unit 740 _{prepares an input value IN i} (0 <i ≦ I; i and I are natural numbers) for the calculation module M. As an example, the selection unit 740 generates I-type input value IN based on the detection result DR among the log data stored in the log data storage unit 80.
(Step S220) The selection unit 740 initializes the module counter n for module determination to n = 1.
(Step S230) The selection unit 740 initializes the input value counter i for module determination to i = 1.

(Step S240) The selection unit 740 _{inputs (supplies) the input value IN i} generated in step S210 to the pre-update calculation module MBFn, and acquires _{the output value OUT i_before} of the pre-update calculation module MBFn obtained as a result. To do. The selection unit 740 stores the acquired output value OUT _{i_before} in a storage unit (not shown).

(Step S250) The selection unit 740 _{inputs (supplies) the input value IN i} generated in step S210 to the post-update calculation module MATn, and acquires _{the output value OUT i_after} of the post-update calculation module MATn obtained as a result. To do. The selection unit 740 stores the acquired output value OUT _{i_after} in a storage unit (not shown).

(Step S260) The selection unit 740 determines whether or not the input value counter i has reached the number of types I of the input value IN. When the selection unit 740 determines that the input value counter i has not reached the number of types I of the input value IN (step S260; NO), the selection unit 740 increments the input value counter i and returns the process to step S240. .. When the selection unit 740 determines that the input value counter i has reached the number of types I of the input values (step S260; YES), the selection unit 740 proceeds to the process in step S270.

(Step S270) selecting section 740 compares the output value _{OUT I_before} before updating operation module MBFn obtained in step S240, an output value _{OUT I_after} obtained updated calculation module MATn in step S250.

FIG. 13 is a diagram showing an example of comparison between the calculation modules M by the selection unit 740 of the present embodiment. Selecting unit 740 compares the output value _{OUT I_before} the pre-update computation module MBF outputs, and an output value _{OUT I_after} the updated calculation module MAT outputs. Selecting unit 740, an output value _{OUT I_before} the pre-update computation module MBF outputs the difference between the output value _{OUT I_after} the updated calculation module MAT is output from the updated calculation module MAT exceeding a predetermined criteria, distribution target Select as the arithmetic module MDT.

Here, the predetermined determination criterion is determined based on the Euclidean distance between the output values OUT as an example. In this case, when _{the Euclidean distance between the output value OUT i_before} and the output value OUT _{i_after} of a certain arithmetic module M is more than a predetermined distance, the selection unit 740 distributes the arithmetic module M to the distribution target arithmetic module. Select as MDT.

The set of the output value OUT _{i_before} and the output value OUT _{i_after} exists in a number corresponding to the number of types I of the input value IN. In this case, the selection unit 740 _{regards the set of the output value OUT i_before} and the output value OUT i_after as a vector, and considers the set to be the average value of the number of sets of the Euclidean distance between the _{output value OUT i_before} and the output value OUT _{i_after (that is,).} Average Euclidean distance) is calculated. The selection unit 740 selects the calculation module M as the distribution target calculation module MDT when the average Euclidean distance calculated for a certain calculation module M is separated by a predetermined distance or more.

(Step S280) Returning to FIG. 12, the selection unit 740 determines whether or not the module counter n has reached the number of modules m of the calculation module M to be selected. When the selection unit 740 determines that the module counter n has not reached the number m of modules of the arithmetic module M (step S280; NO), the module counter n is incremented and the process is returned to step S230. When the selection unit 740 determines that the module counter n has reached the number of modules m of the arithmetic module M (step S280; YES), the selection unit 740 ends the process.

[Summary of Embodiment]
As described above, the distribution device 70 of the present embodiment selects and distributes the distribution target calculation module MDT from the updated calculation module MAT stored in the module library 90. According to the distribution device 70 configured in this way, the amount of communication at the time of distribution of the arithmetic module M is reduced as compared with the case where all the updated arithmetic modules MAT stored in the module library 90 are distributed, for example. be able to.

Further, in the distribution device 70 of the present embodiment, the selection unit 740 selects the updated calculation module MAT in which the number of learnings for the calculation module M exceeds a predetermined threshold value as the distribution target calculation module MDT. Here, if the number of learnings exceeds a predetermined value, the learning result of the calculation module M is saturated to a predetermined degree, and the output value OUT becomes more stable. According to the distribution device 70 configured in this way, it is possible to distribute the arithmetic module M that is well learned and the learning result is more stable.

Further, in the distribution device 70 of the present embodiment, the selection unit 740 determines the difference between the output value OUT output by the pre-update calculation module MBF and the output value OUT output by the post-update calculation module MAT in the selection unit 740. The updated arithmetic module MAT that exceeds the judgment criteria of is selected as the distribution target arithmetic module MDT. Here, if the difference between the output value OUT before learning and the output value OUT after learning of the calculation module M is large, it indicates that the calculation module M has been well learned. According to the distribution device 70 configured in this way, the well-learned arithmetic module M can be distributed.

Further, by configuring the arithmetic module M as described above, the distribution device 70 can be trained for each arithmetic module M, and the other arithmetic modules M can be trained by using the learned arithmetic module M. Can be done. Therefore, according to the distribution device 70, the learning efficiency is higher than in the case where all the arithmetic modules M are learned at the same time, and the learning can be performed in a short period of time. That is, according to the distribution device 70, it is possible to reduce the difficulty of learning the calculation module M for controlling the behavior of the vehicle CR.

The selection unit 740 is a second selection condition based on the comparison result between the first selection condition based on the number of learnings for the calculation module M described above and the output value of the pre-update calculation module MBF and the output value of the post-update calculation module MAT. The distribution target calculation module MDT may be selected in combination with the selection condition.
Specifically, the selection unit 740 selects the calculation module M whose learning count exceeds a predetermined threshold value as a distribution candidate, and distributes by comparing the output values from the calculation modules M selected as the distribution candidates. Select the target calculation module MDT. That is, the selection unit 740 selects the distribution target calculation module MDT based on the second selection condition from the calculation modules M narrowed down based on the first selection condition. Here, the processing load of the selection process based on the second selection condition (for example, the comparison result between the output values of the calculation module M) is larger than the processing load of the selection process based on the first selection condition (for example, the number of learnings). Is also big. That is, the selection unit 740 makes a plurality of stages of determination, with the selection condition having a relatively small processing load as the first selection condition and the selection condition having a relatively large processing load as the second selection condition.
With this configuration, the distribution device 70 of the present embodiment can reduce the processing load of the selection unit 740 as compared with the case where the determination is made only by the selection conditions having a relatively large processing load.

[Modification example (1) Selection by comparing output values of arithmetic modules]
In the above-described embodiment, it has been described that one arithmetic module M is compared with the output value OUT of the pre-update arithmetic module MBF and the output value OUT of the post-update arithmetic module MAT. In this modification, the output value OUT of the arithmetic module group before the update and the output value OUT of the arithmetic module group after the update are compared with each other in the arithmetic module group in which a plurality of arithmetic modules M are combined. It is different from the embodiment. Here, the updated arithmetic module group refers to one (or a plurality) arithmetic modules M to be updated among the arithmetic modules M included in the arithmetic module group before the update, and then the update target. This is an arithmetic module group in which the arithmetic module M is replaced with the corresponding arithmetic module M of the arithmetic module group before the update.

FIG. 14 is a diagram showing an example of comparison between arithmetic module groups by the selection unit 740 in the modified example of the present embodiment. In this example, the calculation module group 10a before the update is composed of the calculation module group of the vehicle-mounted control device 10 of the vehicle CR. Here, the calculation module group of the vehicle-mounted control device 10 of the vehicle CR includes the selection module 101, the selection module 102, the selection module 103, the control module 111, the control module 112, the control module 113, the control module 114, the control module 115, and the control. Module 116 is included (see Figure 1).

In this case, the arithmetic module group 10a before the update includes the selection module 101, the selection module 102, the selection module 103, the control module 111, the control module 112, the control module 113, the control module 114, the control module 115, and the control module 116.
Here, a case where the arithmetic module M to be updated is the control module 111 will be described. That is, in this case, the control module 111 corresponds to the pre-update calculation module MBFn.

In this example, the updated arithmetic module group 10b includes a selection module 101, a selection module 102, a selection module 103, a control module 111, a control module 112, a control module 113, a control module 114, a control module 115, and a control module 116. ..
In this example, learning about the control module 111 is performed. The control module 111 corresponds to the updated calculation module MATn.

In this modification, the selection unit 740 sets the distribution target arithmetic module MDT based on the comparison result between the _{output value OUT i_before} of the arithmetic module group 10a before the update and the output value OUT _{i_after} of the arithmetic module group 10b after the update. select. That is, the selection unit 740 has an output value OUT output by a plurality of layered arithmetic modules M including the pre-update arithmetic module MBF, and a plurality of arithmetic modules in which the pre-update arithmetic module MBF is replaced with the post-update arithmetic module MAT. The distribution target calculation module MDT is selected based on the difference from the output value OUT output by M.

The distribution device 70 of this modification selects the distribution target arithmetic module MDT based on not only the output result of the single arithmetic module M but also the output result of the layered arithmetic module M. According to the distribution device 70 configured in this way, it is possible to compare the output results closer to the behavior of the vehicle CR as compared with the case of comparing the output results of a single arithmetic module M. That is, according to the distribution device 70 configured in this way, the well-learned arithmetic module M can be distributed in an environment closer to the behavior of the vehicle CR.

[Selection based on the hierarchy of arithmetic modules]
As described above with reference to FIG. 8, as a result of learning of the arithmetic module M from the lower layer to the upper layer, the updated arithmetic module MAT may exist in a plurality of layers. In this case, the selection unit 740 selects the distribution target calculation module MDT for each layer.

Specifically, the selection unit 740 selects the updated arithmetic module MAT existing in the lower layer of the hierarchy as the distribution target arithmetic module MDT, and then selects the updated arithmetic module MAT existing in the upper layer of the hierarchy as the distribution target arithmetic. Select as module MDT.

A more specific description will be given with reference to FIGS. 15 and 16.
FIG. 15 is a diagram showing an example of a first-stage comparison between arithmetic module groups by the selection unit 740 in the modified example of the present embodiment.
Here, the case where the learning unit 730 learns the control module 111 and the control module 115 will be described as an example. In this case, the control module 111 is the upper layer arithmetic module M, and the control module 115 is the lower layer arithmetic module M. The selection unit 740 determines whether or not to select the control module 111 and the control module 115 as the distribution target calculation module MDT (that is, distribution necessity).
In the case of this example, the calculation module group 10c includes a control module 111 before update (pre-update calculation module MBF ₁₁₁ ) and a control module 115 before update (pre-update calculation module MBF ₁₁₅ ).

In the first stage, the selection unit 740 performs an output comparison when the lower layer arithmetic module M (in this example, the control module 115) is replaced. _{More specifically, the selection unit 740 replaces the pre-update arithmetic module MBF 115} included in the arithmetic module group 10c with the post-update arithmetic module MAT ₁₁₅ to generate the arithmetic module group 10d. Selecting unit 740 compares the output value _{OUT I_before} calculation module group 10c, and an output value _{OUT I_after} computing modules 10d.

Selecting unit 740, based on a result of comparison between the output value _{OUT I_before} the output value _{OUT I_after,} determines delivery necessity of update after computation module _{MAT 115.} Here, a case where the selection unit 740 selects the updated calculation module MAT ₁₁₅ as a distribution target will be described as an example. When the lower layer arithmetic module M is selected as the distribution target, the selection unit 740 also determines whether or not the upper layer arithmetic module M of the arithmetic module M needs to be distributed. In this example, when the update calculation module MAT ₁₁₅ is selected as the distribution target, the selection unit 740 determines whether or not the _{update calculation module MAT 111} , which is the upper layer calculation module M, needs to be distributed.

FIG. 16 is a diagram showing an example of a second stage comparison between the calculation module group by the selection unit 740 in the modified example of the present embodiment.
In the second stage, the selection unit 740 performs an output comparison when the upper layer arithmetic module M (in this example, the control module 111) is replaced. More specifically, the selection unit 740 replaces the upper control module 111 (pre-update arithmetic module MBF ₁₁₁ ) included in the arithmetic module group 10d with the post-update arithmetic module MAT ₁₁₁ to generate the arithmetic module group 10e. .. Selecting unit 740 compares the output value _{OUT I_before} computing modules 10d, and an output value _{OUT I_after} computing modules 10e.

Selecting unit 740, based on a result of comparison between the output value _{OUT I_before} the output value _{OUT I_after,} determines delivery necessity of update after computation module _{MAT 111.}

In this example, the selection unit 740 determines whether or not the lower layer arithmetic module M needs to be distributed, and then determines whether or not the upper layer arithmetic module M needs to be distributed, but the present invention is not limited to this. The selection unit 740 may determine whether or not the upper layer arithmetic module M needs to be distributed, and then determine whether or not the lower layer arithmetic module M needs to be distributed.

As described above, the selection unit 740 of this modification determines whether or not the calculation module M needs to be distributed for each layer of the calculation module M. According to the distribution device 70 configured in this way, a plurality of arithmetic modules M can be distributed together. Further, in the selection unit 740 of this modification, when a plurality of arithmetic modules M are collectively distributed, it is necessary to replace the arithmetic module M of the upper layer (or lower layer) by replacing the arithmetic module M of the lower layer (or upper layer). To judge. That is, the selection unit 740 of this modification determines whether or not distribution of another calculation module M due to the replacement of the calculation module M is necessary in consideration of the influence between layers.
Here, in the case of distribution in consideration of the influence between the layers of the arithmetic module M, it is conceivable to distribute all the arithmetic modules M located in the upper layer of the arithmetic module M together with the arithmetic module M to be distributed. However, when all the arithmetic modules M located above the arithmetic module M to be distributed are distributed, there arises a problem that the communication amount increases according to the number of the arithmetic modules M to be distributed.
According to the distribution device 70 of this modification, in order to determine whether or not the upper layer arithmetic module M needs to be distributed, communication is performed as compared with the case where all the arithmetic modules M located in the upper layer of the arithmetic module M to be distributed are distributed. The amount can be reduced.

[Modification example (2) Deliver according to the type of vehicle CR]
The distribution priority of the updated calculation module M may be determined by the type of vehicle CR. Here, the updated arithmetic module M (post-update arithmetic module MAT) can more appropriately control the behavior of the vehicle CR as compared with the pre-update arithmetic module M (pre-update arithmetic module MBF). In this case, the type of vehicle CR used for public transportation (for example, buses and taxis) and the type of vehicle CR (for example, heavy machinery) that have a relatively large impact on road equipment, surrounding environment, etc. For vehicles, heavy trucks, etc.), the distribution priority of the arithmetic module M may be higher than that of other types of vehicle CRs.

In this case, the selection unit 740 (delivery destination vehicle selection unit 742) selects from a plurality of distribution target calculation modules MDT according to the type of vehicle CR based on the distribution priority according to the type of vehicle CR. .. Further, the distribution device communication unit 710 distributes the distribution target calculation module MDT to the vehicle CR of the selected type.

According to the distribution device 70 configured in this way, the calculation module M is sequentially distributed for each type of vehicle CR according to the update priority of the calculation module M. Therefore, according to the distribution device 70, it is possible to reduce the amount of communication per hour for distribution of the calculation module M as compared with the case where the calculation module M is collectively distributed to all types of vehicle CRs. ..

Although the embodiments of the present invention and modifications thereof have been described above, these embodiments and modifications thereof are presented as examples, and are not intended to limit the scope of the invention. These embodiments and modifications thereof can be implemented in various other embodiments, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and at the same time, are included in the scope of the invention described in the claims and the equivalent scope thereof.

Each of the above-mentioned devices has a computer inside. The process of each process of each device described above is stored in a computer-readable recording medium in the form of a program, and the process is performed by the computer reading and executing this program. Here, the computer-readable recording medium refers to a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Further, this computer program may be distributed to a computer via a communication line, and the computer receiving the distribution may execute the program.

Further, the above program may be for realizing a part of the above-mentioned functions.
Further, it may be a so-called difference file (difference program) that can realize the above-mentioned function in combination with a program already recorded in the computer system.

10 ... In-vehicle control device, 100 ... Selection module, 110 ... Control module, 120 ... Acquisition unit, 130 ... Output unit, 20 ... Detection information generation unit, 30 ... In-vehicle device, 40 ... Logger, 50 ... Communication unit, 60 ... Module Update unit, 70 ... Distribution device, 710 ... Distribution device communication unit, 720 ... Information acquisition unit, 730 ... Learning unit, 740 ... Selection unit, 80 ... Log data storage unit, 90 ... Module library, CMD ... Control information, DR ... Detection result, M ... Calculation module

Claims (7)

The behavior of the vehicle is controlled by generating vehicle control information based on a plurality of layered arithmetic modules including a neuro arithmetic module that performs arithmetic by a neural network and a detection result in which the situation around the vehicle is detected. A distribution device that distributes the arithmetic module to an in-vehicle control device.
An information acquisition unit that acquires the detection result and the control information from a storage unit that stores the detection result and the control information acquired from the vehicle.
A learning unit that learns about the neuro calculation module based on the detection result and the control information acquired by the information acquisition unit.
Among the updated calculation modules that are the neuro-calculation modules updated by the learning performed by the learning unit, the selection unit that selects the updated calculation module that satisfies a predetermined selection condition as the distribution target calculation module.
A distribution unit that distributes the distribution target calculation module selected by the selection unit to the in-vehicle control device, and
Distributor equipped with. The predetermined selection condition is a condition based on the number of times of learning for the calculation module.
The distribution device according to claim 1, wherein the selection unit selects the updated calculation module whose number of learnings for the calculation module exceeds a predetermined threshold value as the distribution target calculation module. The predetermined selection condition is a condition based on a comparison between the output value output by the pre-update calculation module, which is the calculation module before the update by learning of the learning unit, and the output value output by the post-update calculation module. ,
The selection unit uses the post-update calculation module whose difference between the output value output by the pre-update calculation module and the output value output by the post-update calculation module exceeds a predetermined determination criterion as the distribution target calculation module. The distribution device according to claim 1 or 2, which is selected. The predetermined selection condition is a condition based on a comparison between the output value output by the pre-update calculation module, which is the calculation module before the update by learning of the learning unit, and the output value output by the post-update calculation module. ,
The selection unit includes output values output by a plurality of layered arithmetic modules including the pre-update arithmetic module, and the plurality of arithmetic modules in which the pre-update arithmetic module is replaced with the post-update arithmetic module. The distribution device according to claim 1 or 2, wherein the distribution target calculation module is selected based on the difference from the output value to be output. The distribution device according to claim 4, wherein the selection unit selects the distribution target calculation module for each of the layers when the updated calculation module exists in a plurality of layers. The selection unit selects the updated arithmetic module existing in the lower layer of the hierarchy as the distribution target arithmetic module, and then uses the updated arithmetic module existing in the upper layer of the hierarchy as the distribution target arithmetic module. The distribution device according to claim 5 to be selected. The delivery target calculation module is associated with the delivery target vehicle type, and is associated with the delivery target calculation module.
From claim 1, the distribution unit distributes the distribution target calculation module selected according to the vehicle type from among the plurality of distribution target calculation modules based on the distribution priority according to the vehicle type. The distribution device according to any one of claims 6.

Images