Control device

Abstract

Provided is an automatic control technique capable of reducing the frequency of user intervention. A control device according to one side of the present disclosure is configured to select one or more control models from a plurality of control models by selecting a model, derive a control command for a moving body using the selected one or more control models, and control an operation of the moving body in accordance with the derived control command without intervening an operation. When there is an interventional operation, the control unit discards the derived control command or superimposes the derived control command, and controls the movement of the moving body in accordance with the interventional operation by the user. The selection model is configured to select one or more control models in accordance with an intervention history of a user to avoid the occurrence of intervention.

Description

Technical Field

[0001] The present disclosure relates to a control device. Background Art

[0002] Patent Document 1 proposes a system for autonomous vehicle control, which is configured to use a trained neural network to determine vehicle commands based on a route, GPS data, and sensor data.

[0003] Prior art literature

[0004] Patent Document 1: Japanese Patent Application No. 2019-533810 Summary of the invention

[0005] Technical problem to be solved by the invention

[0006] One of the objectives of the present disclosure is to provide an automatic control technology that can reduce the frequency of user intervention.

[0007] Technical solutions to solve problems

[0008] The control device involved in the first technical solution of the present disclosure includes a storage unit and a control unit for storing multiple control models and selection models. Each control model is configured to derive control instructions for automatically controlling the movement of a mobile body. The control unit is configured to execute: selecting one or more control models from the multiple control models through the selection model; deriving control instructions for the mobile body using the selected one or more control models; controlling the movement of the mobile body according to the derived control instructions in the absence of user intervention; and in the presence of user intervention, discarding the derived control instructions or superimposing them on the derived control instructions, and controlling the movement of the mobile body according to the user intervention. In addition, the selection model is configured to select the one or more control models based on the user's intervention history so as to avoid the user's intervention. In addition, at least any one of the multiple control models and the selection model can use a trained machine learning model. The machine learning model can use a neural network.

[0009] Effects of the Invention

[0010] According to the present disclosure, it is expected that the frequency of user intervention can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 An example of a scenario to which the present disclosure is applied is shown in a schematic manner.

[0012] Figure 2An example of a data structure of an intervention history record of the present disclosure is shown in a schematic manner.

[0013] Figure 3 An example of the hardware configuration of the control device of the present disclosure is schematically shown.

[0014] Figure 4 An example of a processing procedure related to control by the control device of the present disclosure is shown.

[0015] Description of Reference Numerals

[0016] 1 control device; 11 control unit; 12 storage unit; 20 selection model; 30, 35 control model; 50 control instruction; 55 operation; 60 intervention history; M moving body; S sensor DETAILED DESCRIPTION

[0017] In the past, rule-based autonomous driving systems have been known. In addition, according to the methods of Patent Document 1 and the like, an autonomous driving system can be constructed by using a trained machine learning model. However, the control of autonomous driving based on a control model (rule-based model or machine learning model) is not necessarily suitable for the user. In the case where the control of autonomous driving based on the control model is not suitable for the user, user intervention will occur, and autonomous driving will not be implemented but manual driving will be implemented. In the case where the autonomous driving based on the control model is significantly unsuitable for the user, the autonomous driving system is hardly used, and user intervention may occur frequently. In addition, this problem may occur regardless of the type of vehicle. In addition, the occurrence of such a problem is not limited to the scenario of controlling the vehicle. In terms of controlling the moving body, the same is true for moving bodies other than the vehicle. Therefore, the same problem may also occur in the scenario of controlling all moving bodies other than the vehicle.

[0018] In contrast, the control device involved in the first technical solution of the present disclosure includes a storage unit for storing multiple control models and a selection model, and a control unit. Each control model is configured to derive control instructions for automatically controlling the movement of a mobile body. The control unit is configured to execute: selecting one or more control models from multiple control models through a selection model; deriving control instructions for the mobile body using the selected one or more control models; controlling the movement of the mobile body according to the derived control instructions in the absence of user intervention; and discarding the derived control instructions or superimposing them on the derived control instructions in the presence of user intervention, and controlling the movement of the mobile body according to the user intervention. The selection model is configured to select one or more control models based on the user's intervention history to avoid user intervention.

[0019] In the first technical solution of the present disclosure, the selection model is constructed to select a control model to avoid user intervention based on an intervention history record representing a record of the user's past intervention operations. In a simple example, the selection model can be constructed so that in a scenario where intervention occurred in the past, in a new opportunity, the control model used in the scenario is not used and other control models are selected. In this way, by using the selection model constructed based on the intervention history record, the control model used for automatic control is adjusted, and the derived control instructions can be made suitable for the user. Therefore, according to the present disclosure, it can be expected to reduce the frequency of user intervention.

[0020] As other forms of the control device involved in the above technical solution, one aspect of the present disclosure may be an information processing method that realizes all or part of the above components, or a program, or a storage medium that stores such a program and can be read by a computer or other device. Here, the storage medium that can be read by a device refers to a medium that stores information such as a program through electrical, magnetic, optical, mechanical or chemical effects.

[0021] [1 Application example]

[0022] Figure 1 An example of a scenario to which the present disclosure is applied is schematically shown. The control device 1 involved in the present embodiment is one or more computers configured to control the automatic movement of the target moving body M. In the present embodiment, the control device 1 is mounted on the moving body M and holds a selection model 20 and a plurality of control models 30. Each control model 30 is configured to derive a control instruction in order to perform automatic control of the movement of the moving body M.

[0023] In the present embodiment, the control device 1 selects one or more control models 35 from a plurality of control models 30 by means of the selection model 20, and uses the selected one or more control models 35 to derive the control instruction 50 for the moving body M. It is assumed that if the derived control instruction 50 is appropriate, the automatic control based on the control instruction 50 is easily accepted, and the user's intervention operation 55 is unlikely to occur, but if the derived control instruction 50 is inappropriate, the user's intervention operation 55 is likely to occur. For example, it is assumed that if the speed of the moving body M in the automatic control based on the control model 35 is suitable for the user, the possibility of acceleration or deceleration intervention is low, but if the speed of the moving body M is slow or fast, the possibility of acceleration or deceleration intervention becomes high.

[0024] In the absence of the user's intervention operation 55, the control device 1 controls the movement of the moving body M according to the derived control command 50. On the other hand, in the presence of the user's intervention operation 55, the control device 1 discards the derived control command 50 or superimposes on the derived control command 50, and controls the movement of the moving body M according to the user's intervention operation 55. In the case where the intervention operation 55 occurs, the control device 1 records information indicating the intervention as an intervention history 60. The more the user's intervention occurs, the more intervention history 60 is stored. The selection model 20 is configured to select one or more control models 35 based on the user's intervention history 60 to avoid the user's intervention.

[0025] In a scene that is the same as or similar to a scene where intervention occurred in the past, when the same automatic control as the scene where intervention occurred is repeatedly performed, it is estimated that the possibility of user intervention occurring again is high. Therefore, in this embodiment, the selection model 20 is constructed to select a control model 35 that avoids user intervention based on the record of the user's past intervention operation (intervention history record 60). That is, in this embodiment, on the basis of equipping a plurality of control models 30, the intervention history record 60 is used to optimize the control model 35 used for automatic control (i.e., selected by the selection model 20) based on the user's past intervention tendency. As a result, the control device 1 can avoid using a control model that is estimated to be unsuitable for the user in the target scene based on the past intervention tendency among the plurality of control models 30, and actively use a control model that is likely to be suitable for the user. As a result, according to this embodiment, it is expected that the frequency of user intervention can be reduced. In addition, by reducing the frequency of user operation, it is possible to suppress the wear of the operating parts of the mobile body M (for example, the steering wheel, the accelerator pedal, the brake pedal, etc. when the mobile body M is a vehicle), and it is expected that the life of the operating parts can be extended.

[0026] (Moving body)

[0027] As long as it can be automatically moved by mechanical control, the type of the moving body M can be appropriately selected according to the embodiment. The moving body M can be, for example, a vehicle, an aircraft, a ship, a robot device, or other movable device. The aircraft can be at least one of an unmanned aircraft such as an unmanned aircraft and a manned aircraft. In one example, Figure 1 As shown, the moving body M may be a vehicle. In this case, it is expected that the frequency of user intervention can be reduced when the vehicle is automatically driven. In addition, when the moving body M is a vehicle, the type of vehicle (number of wheels, power source, size, etc.) can be arbitrarily selected. As a typical example, the moving body M may be a car with a level 2 or higher autonomous driving capability.

[0028] (Control the action)

[0029] In one example, the motion of the object moving body M can be controlled by directly controlling the object moving body M. In another example, the moving body M can be provided with a dedicated control device such as a controller. In this case, the motion of the object moving body M can be controlled by the control device 1 by providing the derived result to the dedicated control device, thereby indirectly controlling the object moving body M. In addition, the control device 1 can be installed in any place. In one example, as Figure 1 As shown, the control device 1 may be mounted on the moving object M. In another example, the control device 1 may be arranged away from the moving object M to control the moving object M in a remote manner.

[0030] (Control Model)

[0031] Each control model 30 is constructed to derive control instructions based on the environment of the mobile body M. The environment is an event observed by at least one of the mobile body M itself and the surroundings. In one example, at least a part of the environment can be observed by one or more sensors S arranged inside or outside the mobile body M. As long as the sensor S can observe any environment in which the mobile body M moves, its type may not be particularly limited and may be appropriately selected according to the implementation method. In one example, the one or more sensors S may include a camera (image sensor), a radar, a LiDAR (Light Detection And Ranging)), a sonar (ultrasonic sensor), an infrared sensor, a GNSS (Global Navigation Satellite System) / GPS (Global Positioning Satellite) module, etc.

[0032] As long as the control instructions can be derived from the environment of the mobile body M, the form of the input and output of each control model 30 can be appropriately selected according to the implementation method. In one example, at least any one of the multiple control models 30 can be configured to derive control instructions from the observation data of the sensor at more than one time point. In another example, at least any one of the control models can be configured to derive control instructions from the recognition results of the surrounding environment. In this case, the control device 1 may also have an analytical model that infers the recognition results of the surrounding environment based on the observation data of the sensor. Alternatively, at least any one of the control models may include an analytical model. The analytical model can be constructed arbitrarily. In one example, the analytical model can be composed of a machine learning model. Other information can be arbitrarily added to the input of at least any one of the control models. For example, at least any one of the control models can be configured to also accept the input of any information such as a set speed, a speed limit, a location, map information, and navigation information.

[0033] Each control model 30 can be composed of at least one of a trained machine learning model and a rule base model. The rule base model is configured to compare the provided input (such as observation data, information representing the environment such as recognition results of the surrounding environment) with the rules, and derive control instructions based on the results of the comparison (according to appropriate rules). The rules can be set manually or at least partially automatically. The machine learning model is configured to have one or more operation parameters that can be adjusted through machine learning. One or more operation parameters are used for the operation of inference as the purpose (in the present disclosure, the derivation of control instructions). Machine learning refers to the use of learning data to adjust (optimize) the value of the operation parameter. The machine learning model can be composed of, for example, a neural network, a support vector machine, a regression model, a decision tree model, etc. The machine learning method can be appropriately selected according to the machine learning model adopted (such as the error back propagation method, etc.).

[0034] As an example, at least any one of the control models may be composed of a neural network. The structure of the neural network may be appropriately determined according to the implementation method, for example, by the number of layers from the input layer to the output layer, the type of each layer, the number of nodes (neurons) contained in each layer, the connection relationship between the nodes of each layer, etc. In one example, the neural network may have a recursive structure. In addition, the neural network may include, for example, any layers such as a fully connected layer, a convolutional layer, a pooling layer, a deconvolutional layer, an unpooling layer, a normalization layer, a dropout layer, and an LSTM (Long short-term memory). The neural network may have any mechanism such as an attention mechanism. The neural network may include any model such as a GNN (Graph neural network), a diffusion model, a generative model (such as a Generative Adversarial Network, a Transformer, etc.). When the neural network is used in the control model, the weight of the connection between the nodes contained in the control model and the threshold of each node are examples of operation parameters. In addition, when a machine learning model is adopted, the control model can be constructed by an end-to-end model.

[0035] Furthermore, at least one of the plurality of control models 30 is configured to be able to derive different control instructions in the same environment or in the same movement mode relative to other control models. Thus, when the automatic control based on any one of the plurality of control models 30 is no longer suitable for the user (intervention occurs), the possibility of performing automatic control suitable for the user can be ensured by using other control models. That is, the possibility of finding a control model 35 suitable for the user can be ensured by selecting the selection model 20.

[0036] In one example, the structures of the trained machine learning model and the rule base model may be different. Therefore, the characteristics of the automatic control of the trained machine learning model and the rule base model may be different. Therefore, the multiple control models 30 may include more than one trained machine learning model and more than one rule base model. In this way, the change of the characteristics of the automatic control can be ensured, and the probability of the existence of a control model 30 suitable for the user can be increased. As a result, it can be expected that the frequency of user intervention can be further reduced.

[0037] In another example, the characteristics of the automatic control of the trained machine learning model may depend on the learning conditions. Therefore, the characteristics of the machine learning model trained with different learning conditions (for example, using different training samples for machine learning, changing the sampling probability, etc.) may be different. In addition, if the structure of the machine learning model is different, the inference results of the machine learning model may also be different. Therefore, the multiple control models 30 may include multiple trained machine learning models with different learning conditions and structures. As a result, the same effects as above can be expected. In the case where the multiple control models 30 include multiple rule base models, the rules of each rule base model may be different.

[0038] In addition, the control required for each scene may be different. Therefore, each control model 30 can be prepared for each scene. In one example, when the moving body M is a vehicle, multiple control models 30 that are different for each scene such as lane change, lane maintenance, emergency stop (EDSS: Emergency Driving Stop System) can be prepared. In this case, the multiple control models 30 can be configured to include at least two or more control models, and the at least two or more control models can be used for automatic control in at least one of the scenes that are the objects of automatic control. The same or repeated scene. That is, the control device 1 can maintain more than two control models for the same or repeated scene. As an example, when the moving body M is a vehicle, the multiple control models 30 may include a first control model and a second control model for the same lane change. At this time, the first control model and the second control model can be configured to derive different control instructions even for the same lane change scene, so that the control contents such as the timing, speed, and steering angle of the lane change are different from each other.

[0039] (Control Instructions)

[0040] The control instruction (control instruction 50) is related to the action of the mobile body M. The composition of the control instruction can be appropriately selected according to the implementation method. In one example, the control instruction can be composed of acceleration, deceleration, steering or a combination thereof. Acceleration and deceleration may include shifting. In the case of including at least any one of acceleration, deceleration and steering, the control instruction can be represented by a path. Accordingly, each control model 30 can be represented as a pass planner. In addition, the control instruction may also include instructions related to the operation of the mobile body M. As an example, when the mobile body M is a vehicle, the control instruction may include vehicle operations such as turn indicators, hazard indicators, horns, communication processing (such as sending data to a center, sending an emergency call, etc.).

[0041] In one example, each control model 30 may be configured to directly output a control instruction. In another example, each control model 30 may be configured to indirectly output a control instruction, and the control instruction may be obtained by performing any information processing (interpretation processing) on ​​the output of each control model 30. The control instruction may be configured to directly represent the control amount (control instruction value, control output amount) of the mobile body M, such as the accelerator control amount, the brake control amount, the steering wheel steering angle, etc. Alternatively, the control instruction may be configured to indirectly represent the control amount of the mobile body M, such as a path, a state after control, etc. In this case, the control amount of the mobile body M can be obtained from the control instruction by performing any information processing.

[0042] (Select more than one control model)

[0043] Selecting one or more control models 35 may be configured by selecting one control model 35 or selecting two or more control models 35. Accordingly, deriving the control instruction 50 for the moving body M may be configured by deriving the control instruction 50 from the selected one control model 35 or by integrating the control instructions respectively obtained from the selected two or more control models 35 to derive the control instruction 50. Integration may be performed by any operation such as summing, averaging, weighted averaging, etc. Thus, the control instruction 50 can be appropriately obtained.

[0044] (Abandon or superimpose)

[0045] Abandonment (ignoring) may mean instant switching from automatic control to manual control, that is, instant switching from control based on control instructions 50 derived from one or more control models 35 to control according to user intervention 55. On the other hand, superposition may mean gradual switching from automatic control to manual control (user control).

[0046] (Select Model)

[0047] The selection model 20 is configured to select the control model 35 that is estimated to have a low probability of causing intervention or not causing intervention based on the intervention history record 60. In one example, the selection rule of the selection model 20 can be adjusted so that in an environment that is the same as or close to the environment where intervention occurred in the past (i.e., the range of the environment where intervention occurred in the past), the control model used at that time is not directly used. Whether the conditions are the same or close can be inferred based on the information of the environment such as the observation data obtained by the sensor S.

[0048] In one example, the strategy for specifying the selection rule may be to avoid the control model that caused the intervention and select an appropriate control model (e.g., randomly, in a predetermined order, etc.). In short, the selection model 20 may be configured to select a control model other than the control model used at that time within the scope of the environment where the intervention occurred in the past. Preferably, the selection model 20 may be configured to select a control model that can derive a control instruction that is the same as or close to the control instruction of the user's intervention operation provided at that time.

[0049] In another example, the strategy for defining the selection rule may refer to arbitrarily selecting two or more control models and deriving the integration ratio of the selected two or more control models. The selection model 20 may be configured to select two or more control models within the scope of the environment in which intervention occurred in the past. The two or more control models may be composed of the control model used at that time and one or more other control models, or two or more other control models other than the control model used at that time. The integration ratio of the two or more control models may be appropriately calculated so as to be compatible with the control instruction based on the user intervention at that time.

[0050] As long as the intervention history 60 can be reflected in the selection of the control model 35, the composition of the selection model 20 can be appropriately determined according to the implementation method. In one example, the selection model 20 can be composed of at least one of a rule base model and a trained machine learning model. In the case of adopting the rule base model, in the selection model 20, the information of the intervention history 60 can be directly reflected in the selection rule. In one example, the selection model 20 can be composed of the information of the intervention history 60 and the basis for setting the selection rule according to the intervention history 60. The selection rule can be set manually or at least partially automatically. In the case of adopting the machine learning model, the information indicating the environment in which the intervention occurred in the past, which is represented by the intervention history 60, can be used as a training sample (input data), and the control model (and integration ratio) selected according to the above selection rule can be used as a label (teacher signal, correct answer data) to perform the training of the selection model 20. In this way, a trained machine learning model (selection model 20) that reflects the selection rule following the above strategy can be generated.

[0051] In addition, the plurality of control models 30 may include a trained machine learning model and a rule base model. Accordingly, selecting one or more control models 35 may include: when the intervention history record 60 includes a record (information) indicating a situation in which user intervention occurred during the implementation of automatic control of the mobile body M using the trained machine learning model, selecting the rule base model under the condition indicated by the record. The condition indicated by the record may be in a range that is the same as or close to the environment in which the intervention occurred. The close range may be arbitrarily specified. Thus, in an environment in which the trained machine learning model is not suitable for the user, the control model used can be switched from the trained machine learning model to the rule base model.

[0052] (Intervention History)

[0053] The intervention history record 60 is composed of information on user interventions (records of interventions that occurred in the past) that occurred during the past use of at least any one of the plurality of control models 30 to implement automatic control of the moving body M. The data format of the intervention history record 60 can be appropriately determined according to the implementation method. The intervention history record 60 can be stored in any database format.

[0054] The items of information stored in the intervention history record 60 are not particularly limited as long as they can form the selection model 20, and can be appropriately selected according to the implementation. In one example, the intervention history record 60 may include the conditions of the environment in which the intervention occurred (location, route, etc.) and information for identifying the control model used in the environment of the conditions (such as an identifier, etc.).

[0055] In addition, the intervention history record 60 may include information that can be used to improve the control model 30 (information for formulating rules or information for generating a data set for machine learning) in addition to information used for forming the selection model 20. In one example, the intervention history record 60 may also include the user's intervention operation or a control instruction based on the operation. In the case where the control model 30 is composed of a machine learning model, the information contained in the intervention history record 60 that represents the conditions of the environment in which the intervention occurred can be used as a training sample, and the user's intervention operation or the control instruction based on the operation can be used as a label to implement machine learning (relearning, additional learning, etc.) for updating the control model 30. The machine learning for updating can be performed by the control device 1 or by a computer other than the control device 1 (such as an external server).

[0056] Figure 2 An example of the data structure of the intervention history 60 according to the present embodiment is shown schematically. Figure 2In the example, the record of the intervention history record 60 includes a timestamp, identification information of the control model used (use model), the environmental conditions in which the intervention occurred, and information indicating the intervention operation. The timestamp indicates the date and time when the intervention occurred. The date and time indicated by the timestamp can be used to delete old records, determine the records reflected in the selection model 20 (when the intervention history within a certain period is reflected in the selection model), etc. In addition, the record of the intervention history record 60 can be generated for each intervention operation. The unit of intervention operation represented by the record can be determined arbitrarily. In one example, one record (sampling of the intervention history record) can be generated for each intervention operation. In another example, one record can also be generated corresponding to multiple interventions.

[0057] (user)

[0058] The user may refer to a specific user or an unspecified user. In one example, during the period when the target user uses the control device 1, only the intervention history 60 of the target user may be reflected in the selection rule of the control model 35. In another example, the intervention history 60 of any user including users other than the target user may be reflected in the selection rule of the control model 35. In addition, when the mobile body M is a vehicle, the user is typically a driver.

[0059] [2 Configuration example]

[0060] Figure 3 An example of a hardware configuration of the control device 1 according to the present embodiment is schematically shown. The control device 1 according to the present embodiment is a computer in which a control unit 11, a storage unit 12, an external interface 13, an input device 14, an output device 15, and a driver 16 are electrically connected.

[0061] The control unit 11 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), etc., and is configured to perform arbitrary information processing based on a program and various data. The control unit 11 (CPU) is an example of a processor resource. The storage unit 12 can be composed of, for example, a hard disk drive, a solid state drive, etc. The storage unit 12 (as well as the RAM, ROM) is an example of a memory resource. In the present embodiment, the storage unit 12 stores various information such as the control program 81, the control model data 300, the selection model data 200, and the intervention history data 600.

[0062] The control program 81 is used to cause the control device 1 to execute information processing related to the control of the moving body M (described later). Figure 4 ) program. The control program 81 includes a series of commands for the information processing. The control model data 300 is configured to represent information related to the control model 30. The selection model data 200 is configured to represent information related to the selection model 20. In one example, at least one of the control model data 300 and the selection model data 200 may include information representing the value of the operation parameter adjusted by machine learning. At least one of the control model data 300 and the selection model data 200 may also include information representing the structure of the machine learning model (such as the structure of a neural network, etc.). In another example, at least one of the control model data 300 and the selection model data 200 may include information representing a rule base model (rules). The intervention history data 600 can be appropriately configured to include information representing the intervention history 60.

[0063] The external interface 13 may be, for example, a USB (Universal Serial Bus) port, a dedicated port, a wireless communication port, etc., and may be configured to be connected to an external device in a wired or wireless manner. In the present embodiment, the control device 1 may be connected to the sensor S via the external interface 13. The input device 14 may be, for example, a device for inputting, such as a mouse or a keyboard. The output device 15 may be, for example, a device for outputting, such as a display or a speaker. The input device 14 and the output device 15 may also be integrally configured, for example, by a touch panel display or the like.

[0064] The drive 16 is a device for reading various information such as programs stored in the storage medium 91. At least any one of the control program 81, the control model data 300, the selection model data 200, and the intervention history data 600 may be stored in the storage medium 91 instead of being stored in the storage unit 12 or stored in the storage unit 12 at the same time. The storage medium 91 is configured to store the information by electrical, magnetic, optical, mechanical or chemical action so that a computer or other device can read various information (stored programs, etc.). The control device 1 can obtain at least any one of the control program 81, the control model data 300, the selection model data 200, and the intervention history data 600 from the storage medium 91. In addition, the storage medium 91 may be a disk-type storage medium such as a CD or a DVD, or a storage medium other than a disk-type storage medium such as a semiconductor memory (for example, a flash memory). The type of the drive 16 can be appropriately selected according to the type of the storage medium 91.

[0065] In addition, regarding the specific hardware structure of the control device 1, the components can be appropriately omitted, replaced, and added according to the implementation method. For example, the control unit 11 may also include multiple hardware processors. The hardware processor can be composed of a microprocessor, an FPGA (field-programmable gate array), a DSP (digital signal processor), an ECU (Electronic Control Unit), a GPU (Graphics Processing Unit), etc. At least any one of the external interface 13, the input device 14, the output device 15, and the driver 16 may also be omitted. In addition to being designed as a computer dedicated to the services provided, the control device 1 can be a general-purpose computer, a terminal device, etc.

[0066] [3 Action Examples]

[0067] Figure 4 An example of a processing step related to the control of the mobile body M performed by the control device 1 involved in this embodiment is shown. The following processing steps are an example of a control method executed by a computer. However, the following processing steps are only an example, and each step can be changed as long as it is feasible. In addition, regarding the following processing steps, steps can be appropriately omitted, replaced, and added according to the embodiment.

[0068] In step S101, the control unit 11 operates as an acquisition unit to acquire the observation data 125 of the sensor S. The control unit 11 may acquire the observation data 125 from the sensor S directly or indirectly. In step S102, the control unit 11 operates as a selection unit to select one or more control models 35 from a plurality of control models 30 by selecting the model 20. In one example, when the selection model 20 includes a rule base model, the control unit 11 may derive the selection result of the control model 35 from the environment represented by at least a part of the observation data 125 according to the rule. In another example, when the selection model 20 includes a trained machine learning model, the control unit 11 may acquire the selection result of the control model 35 by inputting at least a part of the observation data 125 into the trained machine learning model and executing the calculation processing of the trained machine learning model. In step S103, the control unit 11 operates as a derivation unit to derive the control instruction 50 of the moving body M from at least a part of the acquired observation data 125 using the selected one or more control models 35. Similar to the selection model 20, in one example, when the control model 35 includes a rule base model, the control unit 11 can derive the control instruction 50 according to the rule. In another example, when the control model 35 includes a trained machine learning model, the control unit 11 can derive the control instruction 50 by executing the calculation processing of the trained machine learning model.

[0069] In addition, in step S102, the control unit 11 may select one control model 35 or more than two control models 35. Accordingly, in step S103, the control unit 11 may derive the control instruction 50 from the selected one control model 35. Alternatively, the control unit 11 derives the control instruction 50 by integrating the control instructions respectively obtained from the selected more than two control models 35. In addition, the order of step S102 and step S103 may not be limited to such an example. In another example, step S103 may also be performed before step S102. In this case, the control unit 11 may derive the control instructions from all control models 30 that have the possibility of being selected, regardless of whether they are selected. Furthermore, the control unit 11 may also extract the control instructions of the control model 35 selected by step S102 from the derived control instructions.

[0070] In step S104, the control unit 11 operates as an intervention accepting unit to determine whether there is an intervention operation 55 by the user. The control unit 11 can accept the user's intervention operation 55 at any time during the period of controlling the movement of the mobile body M. When the intervention operation 55 is accepted (that is, there is an intervention operation 55), the control unit 11 advances the process to step S106. On the other hand, when the intervention operation 55 is not accepted (there is no intervention operation 55), the control unit 11 advances the process to step S105.

[0071] In step S105, the control unit 11 operates as a motion control unit, and controls the motion of the moving body M according to the derived control command 50. When the motion control is completed, the control unit 11 advances the processing to the next step S108.

[0072] In step S106, the control unit 11 operates as an action control unit, discards the derived control instruction 50 or superimposes the derived control instruction 50, and controls the action of the mobile body M according to the user's intervention operation 55. In step S107, the control unit 11 operates as a history generation unit, generates information indicating the user's intervention operation 55, and saves the generated information as an intervention history 60 in a predetermined storage area. The predetermined storage area (save destination) can be selected arbitrarily. In one example, the predetermined storage area can be a RAM, a storage unit 12, a storage medium 91, etc. In the case where the control device 1 is configured to be able to communicate, the predetermined storage area can also be an external computer. In addition, the processing timing of step S107 is not limited to such an example, and can be appropriately changed. When the recording of the intervention history 60 (updating of the intervention history data 600) is completed, the control unit 11 causes the processing to enter the next step S108.

[0073] In step S108, the control unit 11 determines whether to terminate the control of the moving body M. The criterion for determination can be set arbitrarily. In one example, the control unit 11 may determine that the control of the moving body M is not to be terminated while the moving body M is in motion, whereas the control of the moving body M is determined to be terminated based on an arbitrary termination instruction (e.g., a termination operation by the user via the input device 14). If it is determined that the control is not to be terminated, the control unit 11 returns to step S101 and executes the process again from step S101. On the other hand, if it is determined that the control is to be terminated, the control unit 11 causes the process to proceed to step S109.

[0074] In step S109, the control unit 11 acts as a model updating unit to update the selection model 20 according to the generated intervention history 60. In one example, the control unit 11 can update the selection model 20 according to the above strategy. Thus, the selection model 20 is configured to select one or more control models 35 according to the user's intervention history 60 to avoid user intervention. Figure 4 In the example of , each time the control ends, step S109 is executed and the selection model 20 is updated. However, the processing timing of step S109 is not limited to such an example, and can be appropriately determined according to the implementation. In another example, the control unit 11 can update the selection model 20 by executing step S109 at regular intervals (for example, every month). When the update of the selection model 20 is completed, the control unit 11 ends the processing steps of the control device 1 involved in this action example.

[0075] [feature]

[0076] In the present embodiment, through the processing of the above-mentioned step S107 and step S109, the selection model 20 is constructed to select the control model 35 based on the record of the user's past intervention operation to avoid the user's intervention. By using the selection model 20 in step S102, the control device 1 can avoid using the control model that is inappropriate for the user in the target scene among the multiple control models 30, which is estimated to be inappropriate for the user based on the past intervention tendency, and can actively use the control model that is likely to be appropriate for the user. As a result, according to the present embodiment, it can be expected that the frequency of user intervention can be reduced.

[0077] [4 Modifications]

[0078] The embodiments of the present disclosure have been described in detail above, but the descriptions so far are merely illustrative of the present disclosure in all aspects. It goes without saying that various improvements or modifications can be made without departing from the scope of the present disclosure. The processes and means described in the present disclosure can be freely combined for implementation as long as no technical contradictions are generated.

Claims

1. A control device comprising: a storage unit storing a plurality of control models and a selection model; and Control Department, Each control model is configured to derive a control instruction for automatically controlling the movement of the moving body. The control unit is configured to execute: selecting one or more control models from the plurality of control models by the selection model; deriving control instructions for the moving object using the selected one or more control models; In the absence of user intervention, the movement of the moving object is controlled according to the derived control instruction; as well as In the case of the user's intervention operation, the derived control instruction is discarded or superimposed on the derived control instruction, and the movement of the moving object is controlled according to the user's intervention operation, The selection model is configured to select the one or more control models based on the intervention history of the user so as to avoid the intervention of the user.

2. The control device according to claim 1, The multiple control models include trained machine learning models and rule base models.

3. The control device according to claim 2, Selecting the one or more control models include: When the intervention history record includes a record indicating a situation in which the user's intervention occurred during the period in which the trained machine learning model is used to implement automatic control of the moving body, the rule base model is selected under the condition indicated by the record.

4. The control device according to claim 1, The selecting of the one or more control models is constituted by selecting one control model, or selecting two or more control models, The control command for the moving object is derived by obtaining the control command from the one control model, or by integrating control commands respectively obtained from the two or more control models to derive the control command.

5. The control device according to any one of claims 1 to 4, The moving object is a vehicle.

Images