Learning device and learning method

Abstract

When the environment in which a moving object moves changes, it becomes impossible to obtain its own position using the existing trained model. [Solution] The learning device comprises an acquisition unit that acquires detection results of the moving environment detected by a preceding moving object among a plurality of moving objects moving along a predetermined path, a learning unit that updates a trained model using the detection results, and a transmission unit that transmits the updated trained model to a subsequent moving object that moves along the path after the preceding moving object among the plurality of moving objects.

Description

Technical Field

[0001] The present disclosure relates to a learning device and a learning method.

Background Art

[0002] A technology for causing a vehicle to travel by autonomous control in a vehicle manufacturing process is known (e.g., Patent Document 1).

Prior Art Literature

Patent Literature

[0003]

Patent Document 1

Summary of Invention

Problem to be Solved by the Invention

[0004] A moving object such as a vehicle acquires its own position using, for example, detection results of the moving environment obtained by a sensor such as an on-vehicle camera and a trained model that has learned the moving environment, and moves autonomously to a destination based on its own position. However, if a change occurs in the moving environment, there is a possibility that the self-position cannot be acquired.

Means for Solving the Problem

[0005] The present disclosure can be implemented in the following modes.

[0006] (1) According to a first aspect of the present disclosure, a learning device is provided. The learning device includes: an acquisition unit configured to acquire a detection result of a moving environment detected by a preceding moving object among a plurality of moving objects that travel along a predetermined route; a learning unit configured to update a trained model using the detection result; and a transmission unit configured to transmit the updated trained model to a following moving object that travels along the route after the preceding moving object among the plurality of moving objects. With this type of learning device, even if a change occurs in the moving environment, the trained model can be updated using the detection results of the moving environment obtained from the preceding moving object, and the updated trained model can be transmitted to the following moving object. As a result, the following moving object can obtain its own position using the updated trained model and the detection results of the moving environment. (2) In the learning device of the above form, the learning unit may update the learned model using the detection result obtained from the preceding moving body that has moved along the path after the occurrence of a predetermined event. This type of learning device allows for updating the trained model at the appropriate time. (3) In the learning device of the above form, the learning unit may update the learned model using the detection results obtained from the preceding moving body that has moved along the path during a predetermined period of time. This type of learning device can suppress the decrease in the accuracy of position estimation by the trained model when the moving environment changes gradually. (4) In the learning device of the above form, the preceding moving body may be configured to detect the moving environment in more detail than the following moving body. This type of learning device can improve the accuracy of position estimation using a pre-trained model. (5) In the learning device of the above form, the plurality of mobile bodies include a plurality of first type mobile bodies and a plurality of second type mobile bodies of a different type from the first type mobile bodies, and the learning device further includes a classification unit that classifies a first detection result, which is a detection result obtained from a first type leading mobile body among the plurality of first type mobile bodies, and a second detection result, which is a detection result obtained from a second type leading mobile body among the plurality of second type mobile bodies, and the learning unit updates a first trained model using the first detection result and updates a second trained model using the second detection result, and the transmission unit may transmit the updated first trained model to a first type following mobile body among the plurality of first type mobile bodies that moves along the path after the first type leading mobile body, and transmit the updated second trained model to a second type following mobile body among the plurality of second type mobile bodies that moves along the path after the second type leading mobile body. This type of learning device allows for updating the trained model for each type of mobile object and transmitting the updated trained model to each mobile object according to its type. (6) A second embodiment of the present disclosure provides a learning method. This learning method comprises the steps of: acquiring detection results of the moving environment detected by a preceding moving body from among a plurality of moving bodies moving along a predetermined path; updating a trained model using the detection results; and transmitting the updated trained model to a following moving body from among the plurality of moving bodies that moves along the path after the preceding moving body. With this learning method, even if there is a change in the moving environment, the trained model can be updated using the detection results of the moving environment obtained from the preceding moving object, and the updated trained model can be transmitted to the following moving object. As a result, the following moving object can obtain its own position using the updated trained model and the detection results of the moving environment. This disclosure can also be implemented in various forms other than learning devices and learning methods. For example, it can be implemented in the form of a system, a vehicle, a server, a computer program, and a recording medium on which the computer program is stored. [Brief explanation of the drawing]

[0007] [Figure 1] An explanatory diagram showing the configuration of the system according to the first embodiment. [Figure 2] An explanatory diagram showing the configuration of the vehicle according to the first embodiment. [Figure 3] An explanatory diagram showing the server configuration of the first embodiment. [Figure 4] A flowchart illustrating the processing procedure for vehicle driving control. [Figure 5] A flowchart illustrating the procedure for updating the trained model in the first embodiment. [Figure 6] This diagram illustrates how the trained model update process of the first embodiment is executed. [Figure 7] An explanatory diagram showing the system configuration of the second embodiment. [Figure 8] An explanatory diagram showing the server configuration of the second embodiment. [Figure 9] A flowchart illustrating the procedure for updating the trained model in the second embodiment. [Figure 10] An explanatory diagram showing the configuration of the vehicle according to the third embodiment. [Modes for carrying out the invention]

[0008] A. First Embodiment: Figure 1 is an explanatory diagram showing the configuration of a system 10 equipped with a learning device in the first embodiment. In this embodiment, the system 10 comprises a plurality of vehicles 100 and a server 200. Although three vehicles 100 are shown in Figure 1, the number of vehicles 100 is not limited to three; it may be two, four or more, etc. In this embodiment, the vehicles 100 correspond to the mobile bodies in this disclosure, and the server 200 corresponds to the learning device in this disclosure.

[0009] In this disclosure, “mobile object” means an object that can move, such as a vehicle or an electric vertical take-off and landing aircraft (so-called flying car). A vehicle may be a wheeled vehicle or a tracked vehicle, such as a passenger car, truck, bus, motorcycle, car, or construction vehicle. Vehicles include electric vehicles (BEVs: Battery Electric Vehicles), gasoline vehicles, hybrid vehicles, and fuel cell vehicles. If the mobile object is not a vehicle, the terms “vehicle” and “car” in this disclosure may be replaced with “mobile object” as appropriate, and the term “driving” may be replaced with “moving” as appropriate.

[0010] Vehicle 100 is configured to operate autonomously. "Autonomous operation" means operation without the operation of a passenger. Operation refers to operations related to at least one of the following: "going," "turning," or "stopping" of vehicle 100. Autonomous operation is achieved by automatic or manual remote control using a device located outside vehicle 100, or by autonomous control of vehicle 100. Vehicle 100 operating autonomously may have passengers on board who do not perform operation. Passengers who do not perform operation include, for example, people simply sitting in the seats of vehicle 100, or people performing tasks other than operation, such as assembly, inspection, or operating switches, while on board vehicle 100. Operation by a passenger is sometimes called "manned operation."

[0011] In this specification, "remote control" includes "fully remote control," in which all operations of the vehicle 100 are completely determined from outside the vehicle 100, and "partial remote control," in which some operations of the vehicle 100 are determined from outside the vehicle 100. Furthermore, "autonomous control" includes "fully autonomous control," in which the vehicle 100 autonomously controls its own operations without receiving any information from external devices, and "partial autonomous control," in which the vehicle 100 autonomously controls its own operations using information received from external devices.

[0012] FIG. 2 is an explanatory diagram showing the configuration of a vehicle 100. In the present embodiment, the vehicle 100 is configured to be able to travel by autonomous control. The vehicle 100 includes a vehicle control device 110, an actuator group 120, a surrounding sensor 130, and a communication device 140. The actuator group 120 includes at least one actuator. In the present embodiment, the actuator group 120 includes an actuator for a driving device that generates propulsive force for the vehicle 100, an actuator for a steering device that changes the traveling direction of the vehicle 100, and an actuator for a braking device that generates braking force for the vehicle 100. Each actuator included in the actuator group 120 is driven under the control of the vehicle control device 110.

[0013] The surrounding sensor 130 detects the traveling environment around the vehicle 100. In the present embodiment, the surrounding sensor 130 is a camera that captures an image of the traveling environment in front of the vehicle 100. The surrounding sensor 130 outputs an image generated by capturing the traveling environment in front of the vehicle 100 as a detection result of the traveling environment. However, the surrounding sensor 130 is not limited to a camera, and may be, for example, LiDAR. When the surrounding sensor 130 is LiDAR, the surrounding sensor 130 outputs a point cloud generated by scanning the traveling environment as a detection result of the traveling environment.

[0014] The vehicle control device 110 controls each part of the vehicle 100. The vehicle control device 110 is configured by a computer including a processor 111, a memory 112, an input / output interface 113, and an internal bus 114. The processor 111, the memory 112, and the input / output interface 113 are bidirectionally communicably connected via the internal bus 114. The actuator group 120, the surrounding sensor 130, and the communication device 140 are connected to the input / output interface 113. In the present embodiment, the communication device 140 communicates with a server 200 via wireless communication.

[0015] The processor 111 functions as a driving control unit 115 by executing a computer program PG1 pre-stored in memory 112. If there is an occupant in the vehicle 100, the driving control unit 115 can drive the vehicle 100 by controlling the actuator group 120 in accordance with the occupant's operations on the vehicle 100. Regardless of whether there is an occupant in the vehicle 100 or not, the driving control unit 115 can acquire the position information of the vehicle 100 using the detection results of the surrounding sensor 130 and the trained model MD stored in memory 112, and drive the vehicle 100 autonomously based on the position information of the vehicle 100. The trained model MD is pre-trained using a training dataset that includes, for example, multiple images generated by imaging the driving environment and the position information of the location where each image was taken. When the trained model MD receives an image which is the detection result of the surrounding sensor 130, it outputs the position information of the vehicle 100 equipped with the surrounding sensor 130. In the following description, the position information of the vehicle 100 will be referred to as vehicle position information.

[0016] Figure 3 is an explanatory diagram showing the configuration of server 200. Server 200 is located outside vehicle 100. Server 200 consists of a computer comprising a processor 201, memory 202, input / output interface 203, and internal bus 204. Server 200 further comprises a communication device 205 for communicating with vehicle 100. The processor 201, memory 202, and input / output interface 203 are connected bidirectionally via the internal bus 204. The communication device 205 is connected to the input / output interface 203.

[0017] The processor 201 functions as the operation management unit 210, acquisition unit 220, learning unit 230, and transmission unit 240 by executing the computer program PG2 pre-stored in memory 202. The operation management unit 210 determines the order in which multiple vehicles 100 will travel. The operation management unit 210 transmits a travel permission signal and a reference route RR pre-stored in memory 202 to the multiple vehicles 100 in the determined order. The reference route RR shows the route that the vehicles 100 should travel. The route is represented by a node indicating the departure point, a node indicating a waypoint, a node indicating the destination, and links connecting each node. However, if the reference route RR is pre-stored in the memory 112 of the vehicle control device 110, the operation management unit 210 does not need to transmit the reference route RR. When a vehicle 100 receives a travel permission signal, it starts driving under autonomous control. The acquisition unit 220 acquires the detection results of the surrounding sensors 130 mounted on the vehicle 100. The learning unit 230 updates the trained model MD stored in the memory 202 using the detection results acquired by the acquisition unit 220. The transmission unit 240 transmits the trained model MD updated by the learning unit 230 to the vehicle 100.

[0018] As shown in Figure 1, in this embodiment, system 10 is used to autonomously control the movement of vehicle 100 in a factory FC where vehicle 100 is manufactured. Factory FC comprises a first location PL1 and a second location PL2. The first location PL1 and the second location PL2 are connected by a track TR on which vehicle 100 can travel. The first location PL1 is, for example, the location where vehicle 100 is assembled. Upon assembly at the first location PL1, vehicle 100 is ready for autonomous control. After assembly at the first location PL1, vehicle 100 moves autonomously along the track TR to the second location PL2. The second location PL2 is, for example, the location where vehicle 100 is inspected. Vehicle 100 that passes inspection at the second location PL2 is then shipped from the factory FC. At least one marker LM is provided on or around the track TR for vehicle 100 to determine its own position. Landmarks LM include, for example, road markings painted on the surface of the track TR, and signs and buildings installed around the track TR. The reference coordinate system of the factory FC is the global coordinate system GC, and any location within the factory FC can be represented by X, Y, Z coordinates in the global coordinate system GC.

[0019] Figure 4 is a flowchart showing the processing procedure for controlling the driving of vehicle 100. In step S11, the driving control unit 115 acquires vehicle position information using the detection results of the surrounding sensor 130 and the trained model MD. The vehicle position information is the position information that forms the basis for generating the driving control signal. In this embodiment, the vehicle position information includes the position and orientation of vehicle 100 in the global coordinate system GC. In this embodiment, the driving control unit 115 captures images of the driving environment using the camera, which is the surrounding sensor 130, and inputs the images obtained from the capture into the trained model MD. The trained model MD detects a marker LM from the image and outputs vehicle position information estimated from the type, position, and size of the marker LM.

[0020] In step S12, the driving control unit 115 determines the next target location to which the vehicle 100 should go. In this embodiment, the target location is represented by X, Y, Z coordinates in the global coordinate system GC. The driving control unit 115 uses the vehicle position information and the reference path RR received from the server 200 to determine the next target location to which the vehicle 100 should go. The driving control unit 115 determines the target location on the reference path RR beyond the current location of the vehicle 100.

[0021] In step S13, the driving control unit 115 generates a driving control signal to drive the vehicle 100 toward the determined target position. In this embodiment, the driving control signal includes the acceleration and steering angle of the vehicle 100 as parameters. The driving control unit 115 calculates the driving speed of the vehicle 100 from the change in the position of the vehicle 100 and compares the calculated driving speed with the target speed. Overall, the driving control unit 115 determines the acceleration so that the vehicle 100 accelerates if the driving speed is lower than the target speed, and determines the acceleration so that the vehicle 100 decelerates if the driving speed is higher than the target speed. Furthermore, if the vehicle 100 is located on the reference path RR, the driving control unit 115 determines the steering angle and acceleration so that the vehicle 100 does not deviate from the reference path RR, and if the vehicle 100 is not located on the reference path RR, in other words, if the vehicle 100 has deviated from the reference path RR, the driving control unit 115 determines the steering angle and acceleration so that the vehicle 100 returns to the reference path RR. In other embodiments, the driving control signal may include the speed of the vehicle 100 as a parameter, either in place of the acceleration of the vehicle 100 or in addition to the acceleration.

[0022] In step S14, the driving control unit 115 controls the actuator group 120 using the generated driving control signal, thereby driving the vehicle 100 at the acceleration and steering angle indicated in the driving control signal. The driving control unit 115 repeats the acquisition of vehicle position information, determination of target position, generation of driving control signal, and control of the actuator group 120 at predetermined intervals.

[0023] Figure 5 is a flowchart showing the procedure for updating the trained model. Figure 6 is an explanatory diagram showing how the trained model update process is executed. In this embodiment, the trained model update process is repeatedly executed by the server 200 at predetermined intervals.

[0024] As shown in Figure 5, in step S110, the acquisition unit 220 of the server 200 determines whether a predetermined event has occurred. In this embodiment, the predetermined event is that construction work is carried out on or around the track TR. When construction work is carried out on or around the track TR, the driving environment changes. Changes in the driving environment include the removal of marker LM, changes in the position or orientation of marker LM, changes in the appearance of marker LM, the replacement of marker LM with another marker, and the marker LM being hidden and no longer visible from the track TR. In these cases, the existing trained model MD may not be able to estimate vehicle position information, or even if it can estimate vehicle position information, the accuracy of the vehicle position information may decrease. The acquisition unit 220 can determine whether construction work is being carried out, for example, using a surveillance camera (not shown) installed at the factory FC. The acquisition unit 220 may also determine whether construction work is being carried out using construction information acquired from an external source via wired or wireless communication. Construction information may be transmitted to the server 200, for example, by the administrator of system 10 or by workers at the factory FC.

[0025] If it is determined in step S110 that no predetermined event has occurred, the server 200 skips steps S110 and beyond and terminates the trained model update process. On the other hand, if it is determined in step S110 that a predetermined event has occurred, the acquisition unit 220 of the server 200 acquires the driving environment detection result RS from the preceding vehicle 100A that traveled on the track TR after the event occurred in step S120. The preceding vehicle 100A is the vehicle 100 that travels on the track TR before the following vehicle 100B. In this embodiment, the preceding vehicle 100A is equipped with a GSNN (Global Navigation Satellite System) receiver 135 (see Figure 6), and is configured to acquire vehicle position information using the GNSS receiver 135 without having to acquire vehicle position information using the driving environment detection result RS obtained from the surrounding sensor 130 and the trained model MD. However, the preceding vehicle 100A may be equipped with an inertial measurement unit instead of the GNSS receiver 135, and may be configured to acquire vehicle position information using the inertial measurement unit. Alternatively, the leading vehicle 100A may be equipped with an inertial measurement unit in addition to the GNSS receiver 135, and may be configured to acquire vehicle position information using the GNSS receiver 135 and the inertial measurement unit. In this embodiment, the operation management unit 210 determines the order in which the multiple vehicles 100 travel so that the leading vehicle 100A is the first among the multiple vehicles 100 to travel on the track TR after the event occurs. If the order in which the multiple vehicles 100 travel so that the leading vehicle 100A is the first among the multiple vehicles 100 to travel on the track TR does not result in the leading vehicle 100A being the first among the multiple vehicles 100, the operation management unit 210 changes the order in which the multiple vehicles 100 travel so that the leading vehicle 100A is the first among the multiple vehicles 100 to travel on the track TR after the event occurs.

[0026] In step S130, the learning unit 230 of the server 200 updates the trained model MD using the detection result RS of the driving environment acquired from the preceding vehicle 100A. Updating the trained model MD includes retraining the trained model MD and additionally training the trained model MD. For example, the learning unit 230 updates the trained model MD by fine-tuning using a training dataset that includes multiple images of the driving environment obtained from the preceding vehicle 100A and vehicle position information corresponding to each image. For the vehicle position information corresponding to each image, vehicle position information acquired using the GNSS receiver 135 can be used.

[0027] In step S140, the transmission unit 240 of the server 200 transmits the updated trained model MD to the following vehicle 100B. The following vehicle 100B is the vehicle 100 that travels on the track TR after the preceding vehicle 100A. After that, the server 200 terminates the trained model update process. The following vehicle 100B, having received the updated trained model MD from the server 200, overwrites the pre-update trained model MD stored in the memory 112 of the vehicle control device 110 mounted on its own vehicle with the updated trained model MD. Here, it is preferable that the surrounding sensor 130 of the preceding vehicle 100A is configured to detect the driving environment in more detail than the surrounding sensor 130 of the following vehicle 100B. Being able to detect the driving environment in more detail includes high resolution if the surrounding sensor 130 is a camera, and high point cloud density if the surrounding sensor 130 is a LiDAR. If the surrounding sensor 130 of the preceding vehicle 100A is configured to detect the driving environment in more detail than the surrounding sensor 130 of the following vehicle 100B, the detection accuracy of the landmark LM by the updated trained model MD can be improved.

[0028] According to the system 10 of this embodiment described above, the server 200 updates the learned model MD using the driving environment detection result RS obtained from the preceding vehicle 100A, and transmits the updated learned model MD to the following vehicle 100B. Therefore, even if the driving environment changes, the following vehicle 100B, which is driving behind the preceding vehicle 100A, can acquire vehicle position information using the updated learned model MD and the driving environment detection result RS obtained from the surrounding sensor 130 mounted on its own vehicle, and can drive to its destination autonomously.

[0029] Furthermore, in this embodiment, when a predetermined event occurs that changes the driving environment, the server 200 updates the trained model MD using the driving environment detection result RS obtained from the preceding vehicle 100A that drove after the event occurred. Therefore, the trained model MD can be updated at an appropriate timing.

[0030] B. Second Embodiment: Figure 7 is an explanatory diagram showing the configuration of system 10 in the second embodiment. Figure 8 is an explanatory diagram showing the configuration of server 200 in the second embodiment. Figure 9 is a flowchart showing the procedure for updating the trained model in the second embodiment. System 10 in the second embodiment differs from the first embodiment in that server 200 is equipped with a classification unit 225 and the content of the trained model update process is different. The other configurations are the same as in the first embodiment unless otherwise specified. In this embodiment, server 200 corresponds to the learning device in this disclosure.

[0031] As shown in Figure 7, in this embodiment, the multiple vehicles 100 include a first-type vehicle 101 and a second-type vehicle 102. The first-type vehicle 101 and the second-type vehicle 102 are of different vehicle types. Different vehicle types include, for example, different types of surrounding sensors 130, different positions of surrounding sensors 130, and different position estimation algorithms. In this embodiment, the first-type vehicle 101 and the second-type vehicle 102 have different types of surrounding sensors 130. The first-type vehicle 101 is equipped with a camera as its surrounding sensor 130. The surrounding sensor 130 of the first-type vehicle 101 outputs an image as the detection result RS of the driving environment. The second-type vehicle 102 is equipped with a LiDAR as its surrounding sensor 130. The surrounding sensor 130 of the second-type vehicle 102 outputs a point cloud as the detection result RS of the driving environment. In the following explanation, the detection result RS output from the surrounding sensor 130 of the first type vehicle 101 will be referred to as the first detection result RS1, and the detection result RS output from the surrounding sensor 130 of the second type vehicle 102 will be referred to as the second detection result RS2. The first trained model MD1, which is a trained model MD for the first type vehicle 101, outputs vehicle position information when the first detection result RS1 is input, and the second trained model MD2, which is a trained model MD for the second type vehicle 102, outputs vehicle position information when the second detection result RS2 is input.

[0032] As shown in Figure 8, in this embodiment, the processor 201 of the server 200 functions as the operation management unit 210, acquisition unit 220, classification unit 225, learning unit 230, and transmission unit 240 by executing a computer program PG2 pre-stored in memory 202. The acquisition unit 220 acquires the driving environment detection results RS1 and RS2 from each vehicle 100. The classification unit 225 classifies the detection results RS1 and RS2 acquired from each vehicle 100 according to the vehicle type of each vehicle 100. The classification unit 225 classifies the detection results RS1 and RS2 acquired from each vehicle 100 into a first detection result RS1 acquired from a first type vehicle 101 and a second detection result RS2 acquired from a second type vehicle 102. The learning unit 230 updates the first trained model MD1 using the first detection result RS1 and updates the second trained model MD2 using the second detection result RS2. The transmitting unit 240 transmits the updated first trained model MD1 to the first vehicle 101 and the updated second trained model MD2 to the second vehicle 102.

[0033] The trained model update process shown in Figure 9 is repeatedly executed by the server 200 at predetermined intervals. In step S210, the acquisition unit 220 of the server 200 acquires the driving environment detection results RS1 and RS2 from each vehicle 100, and the classification unit 225 classifies the detection results RS1 and RS2 acquired by the acquisition unit 220 according to the vehicle type and stores them in the memory 202.

[0034] In step S220, the learning unit 230 of the server 200 determines whether a predetermined period has elapsed. In this embodiment, the predetermined period is one month. However, the predetermined period may be shorter or longer than one month. If it is determined in step S220 that the predetermined period has not elapsed, the server 200 returns to step S210 and repeats the process from step S210 to step S220 until it is determined in step S220 that the predetermined period has elapsed. In the following description, a first-class vehicle 101 that traveled on the track TR during the above period will be referred to as a first-class preceding vehicle 101A, and a first-class vehicle 101 that travels on the track TR after the above period will be referred to as a first-class following vehicle 101B. A Type 2 vehicle 102 that traveled on the TR track during the above period will be referred to as a Type 2 leading vehicle 102A, and a Type 2 vehicle 102 that travels on the TR track after the above period will be referred to as a Type 2 following vehicle 102B.

[0035] If it is determined in step S220 that a predetermined period has elapsed, the learning unit 230 of the server 200 updates the first trained model MD1 using the first detection result RS1 obtained from the first type preceding vehicle 101A, and updates the second trained model MD2 using the second detection result RS2 obtained from the second type preceding vehicle 102A. In step S240, the transmission unit 240 of the server 200 transmits the updated first trained model MD1 to the first type following vehicle 101B, and transmits the updated second trained model MD2 to the second type following vehicle 102B. After that, the server 200 terminates the trained model update process.

[0036] According to the system 10 of this embodiment described above, when a predetermined period of time has elapsed, the server 200 updates the trained models MD1 and MD2 using the detection results RS1 and RS2 obtained from the preceding vehicles 101A and 102A that traveled during that period, and transmits the updated trained models MD1 and MD2 to the following vehicles 101B and 102B. Therefore, when the driving environment changes gradually, it is possible to suppress a decrease in the accuracy of position estimation by the trained models MD1 and MD2.

[0037] Furthermore, in this embodiment, the classification unit 225 of the server 200 classifies the detection results RS1 and RS2 acquired from each vehicle 100 according to vehicle type. Therefore, the trained models MD1 and MD2 for each vehicle type can be updated using the classified detection results RS1 and RS2.

[0038] C. Third Embodiment: Figure 10 is an explanatory diagram showing the configuration of the lead vehicle 100A in the system 10 of the third embodiment. The third embodiment differs from the first embodiment in that the update of the learned model MD is performed on the lead vehicle 100A rather than on the server 200. The other configurations are the same as in the first embodiment unless otherwise specified. In this embodiment, the vehicle control device 110 mounted on the lead vehicle 100A corresponds to the learning device in this disclosure.

[0039] In this embodiment, the leading vehicle 100A can communicate with the following vehicle 100B via wireless communication using the communication device 140. The processor 111 of the vehicle control device 110 mounted on the leading vehicle 100A functions as a driving control unit 115, an acquisition unit 116, a learning unit 117, and a transmission unit 118 by executing a computer program PG1 pre-stored in the memory 112. The driving control unit 115 acquires the vehicle's position information using the detection results of the surrounding sensor 130 and the learned model MD stored in the memory 112, and can drive the vehicle based on the vehicle's position information. The acquisition unit 116 acquires the detection results of the surrounding sensor 130 mounted on the vehicle. The learning unit 117 updates the learned model MD using the detection results acquired by the acquisition unit 116. The transmission unit 118 transmits the learned model MD updated by the learning unit 117 to the following vehicle 100B. When a following vehicle 100B receives the updated learned model MD from the preceding vehicle 100A, it overwrites the previously updated learned model MD stored in the memory 112 of the vehicle control device 110 mounted on its own vehicle with the updated learned model MD. Note that the following vehicle 100B does not necessarily need to be equipped with an acquisition unit 116, a learning unit 117, and a transmission unit 118.

[0040] As described above, the system 10 in this embodiment allows for updating the learned model MD on the preceding vehicle 100A and transmitting the updated learned model MD to the following vehicle 100B.

[0041] D. Other embodiments: (D1) In the first embodiment, the predetermined events may include, in addition to or instead of construction work, at least one of the following: a change in the weather at the factory location and a change in the location estimation algorithm due to a software update of the vehicle 100. In these cases as well, the existing trained model MD may not be able to obtain correct location information. The server 200 can detect changes in the weather using surveillance cameras installed at the factory FC or weather information obtained from an external source.

[0042] (D2) In the second embodiment, the plurality of vehicles 100 may not include either the first type vehicle 101 or the second type vehicle 102. In this case, the server 200 may not be equipped with a classification unit 225.

[0043] (D3) In each of the above embodiments, the vehicle 100 only needs to have a configuration that allows it to move by unmanned operation, and may take the form of a platform having the configuration described below. Specifically, in order for the vehicle 100 to perform the three functions of "driving," "turning," and "stopping" by unmanned operation, it only needs to be equipped with at least a vehicle control device 110, an actuator group 120, and a surrounding sensor 130. When the vehicle 100 acquires information from the outside for unmanned operation, the vehicle 100 may further be equipped with a communication device 140. That is, the vehicle 100 that can move by unmanned operation does not need to have at least some of the interior parts such as the driver's seat and dashboard installed, at least some of the exterior parts such as the bumper and fender installed, and does not need to have a body shell installed. In this case, the remaining parts such as the body shell may be attached to the vehicle 100 before it is shipped from the factory FC, or the remaining parts such as the body shell may be attached to the vehicle 100 after it has been shipped from the factory FC without the remaining parts such as the body shell being attached to the vehicle 100. Each part may be attached to the vehicle 100 from any direction, such as the top, bottom, front, rear, right, or left side, and each part may be attached from the same direction or from different directions.

[0044] (D4) Vehicle 100 may be manufactured by combining multiple modules. A module means a unit composed of one or more parts grouped together according to the configuration and function of vehicle 100. For example, the platform of vehicle 100 may be manufactured by combining a front module that constitutes the front part of the platform, a central module that constitutes the central part of the platform, and a rear module that constitutes the rear part of the platform. The number of modules that constitute the platform is not limited to three, but may be two or fewer, or four or more. In addition to the platform, or in place of the platform, parts of vehicle 100 other than the platform may be modularized. Various modules may also include any exterior parts such as bumpers and grilles, or any interior parts such as seats and consoles. Furthermore, not limited to vehicle 100, any type of mobile body may be manufactured by combining multiple modules. Such modules may be manufactured, for example, by joining multiple parts by welding or fasteners, or by integrally molding at least a part of the module as a single part by casting. The molding method of integrally molding at least a part of the module as a single part is also called gigacast or megacast. By using Gigacast, parts of a mobile body that were conventionally formed by joining multiple components can be formed as single components. For example, the front module, central module, and rear module mentioned above may be manufactured using Gigacast.

[0045] (D5) Transporting vehicle 100 using the unmanned operation of vehicle 100 is also called "autonomous transport." The configuration for realizing autonomous transport is also called a "vehicle remote control autonomous driving transport system." Furthermore, a production method that uses autonomous transport to produce vehicle 100 is also called "autonomous production." In autonomous production, for example, at a factory fuel cell (FC) that manufactures vehicle 100, at least a portion of the transport of vehicle 100 is realized by autonomous transport.

[0046] This disclosure is not limited to the embodiments described above, and can be implemented in various configurations without departing from its spirit. For example, the technical features in the embodiments corresponding to the technical features in each form described in the summary of the invention can be replaced or combined as appropriate in order to solve some or all of the above-described problems, or to achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be deleted as appropriate. [Explanation of Symbols]

[0047] 10...System, 100...Vehicle (Mobile Unit), 101...Type 1 Vehicle (Type 1 Mobile Unit), 102...Type 2 Vehicle (Type 2 Mobile Unit), 110...Vehicle Control Device, 111...Processor, 112...Memory, 113...Input / Output Interface, 114...Internal Bus, 115...Driving Control Unit, 116...Acquisition Unit, 117...Learning Unit, 118...Transmission Unit, 120...Actuator Group, 130...Surrounding Sensor, 135...GNSS Receiver, 140...Communication Device, 200...Server, 201...Processor, 202...Memory, 203...Input / Output Interface, 204...Internal Bus, 205...Communication Device, 210...Operation Management Unit, 220...Acquisition Unit, 225...Classification Unit, 230...Learning Unit, 240...Transmission Unit

Claims

1. A learning device, An acquisition unit that acquires the detection results of the movement environment detected by a preceding moving object among multiple moving objects moving along a predetermined route, A learning unit that updates the trained model using the detection results, A transmitting unit that transmits the updated trained model to a subsequent mobile body among the plurality of mobile bodies that moves along the path after the preceding mobile body, A learning device equipped with the following features.

2. A learning device according to claim 1, The learning unit is a learning device that, when a predetermined event occurs, updates the trained model using the detection results obtained from the preceding moving object that moved along the path after the event occurred.

3. A learning device according to claim 1, The learning unit is a learning device that updates the learned model using the detection results obtained from the preceding moving object that moved along the path during a predetermined period of time.

4. A learning device according to claim 1, The preceding moving body is configured to be able to detect the moving environment in more detail than the following moving body, and is a learning device.

5. A learning device according to claim 1, The aforementioned plurality of mobile bodies include a plurality of first-type mobile bodies and a plurality of second-type mobile bodies that are of a different type from the first-type mobile bodies. The learning device further includes a classification unit that classifies a first detection result, which is a detection result obtained from a first type preceding mobile body among the plurality of first type mobile bodies, and a second detection result, which is a detection result obtained from a second type preceding mobile body among the plurality of second type mobile bodies. The learning unit updates the first trained model using the first detection result, and updates the second trained model using the second detection result. A learning device comprising: a transmitting unit that transmits the updated first trained model to a first type following mobile unit among the plurality of first type mobile units that moves along the path after the first type preceding mobile unit; and a transmitting unit that transmits the updated second trained model to a second type following mobile unit among the plurality of second type mobile units that moves along the path after the second type preceding mobile unit.

6. It is a learning method, A step of obtaining the detection results of the movement environment detected by a preceding moving object among multiple moving objects moving along a predetermined route, The process of updating the trained model using the detection results, The process of transmitting the updated trained model to a subsequent mobile body among the plurality of mobile bodies that moves along the path after the preceding mobile body, A learning method that includes [the following features].

Images