Vehicle control device, vehicle control program and vehicle control method

Abstract

To provide a vehicle control device, a vehicle control program and a vehicle control method which can suppress an influence by change of a travel state of a vehicle.SOLUTION: A motor controller 60 for controlling motors 40a and 40b of a vehicle 100 includes: a measurement part 61 for estimating a sideslip angle β, on the basis of a measurement result of an on-vehicle sensor 70; an input part 62 for inputting an angular velocity instruction value ω*; a drive control part 63 for controlling the motors 40a and 40b on the basis of the angular velocity instruction value ω*; a prediction part 64 for inputting the angular velocity instruction value ω* input to the input part 62 to a prediction model, and predicting a sideslip angle β' by the prediction model; and a correction part 65 for comparing the sideslip angle β estimated by the measurement part 61 with the sideslip angle β' predicted by the prediction part 64, performing the feedback of a response difference Δβ and correcting the angular velocity instruction value ω* input to the input part 62.SELECTED DRAWING: Figure 6

Description

[Technical Field]

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

[0002] Conventionally, there has been known a yaw moment control device that controls the yaw moment by varying the ratio of driving force distributed to the left and right drive wheels and also controls the yaw moment by distributing braking force to the left and right wheels in order to prevent the vehicle from skidding. Such a yaw moment control device can improve the stability of the vehicle. Such a control device is described, for example, in Patent Document 1. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2011-162145 Summary of the Invention [Problem to be solved by the invention]

[0004] When calculating parameters necessary for control, such as yaw rate, it is common to use a reference vehicle model. However, since it is desirable for the vehicle model to be changed in response to changes in road surface and tire conditions, there has been an issue with robustness.

[0005] The present invention has been made in consideration of the above circumstances, and its main object is to provide a vehicle control device, a vehicle control program, and a vehicle control method that can suppress the effects of changes in the vehicle's driving conditions. [Means for solving the problem]

[0006] A first means for solving the above problem is a vehicle control device that controls a power source of a vehicle, and includes: a measurement unit that estimates at least one of driving parameters correlated with the driving state of the vehicle based on measurement results from on-board sensors; an input unit that inputs a control command value for controlling the power source; a drive control unit that controls the power source based on the control command value to drive the vehicle; a prediction unit that inputs the control command value input to the input unit into a prediction model and predicts the driving parameters using the prediction model; and a correction unit that compares the driving parameters estimated by the measurement unit with the driving parameters predicted by the prediction unit, feeds back the difference between them so that the estimated driving parameters approach the predicted driving parameters, corrects the control command value input to the input unit, and inputs it to the drive control unit.

[0007] According to the above means, the control command value can be corrected so that the driving parameters estimated by the measurement unit approach the predicted driving parameters, thereby suppressing the influence of changes in the vehicle's driving state.

[0008] A second means for solving the above problem is a vehicle control program executed by a vehicle control device that controls a power source of a vehicle, which causes the vehicle control device to carry out the following steps: a measurement step of estimating at least one of driving parameters correlated with the driving state of the vehicle based on measurement results of on-board sensors; an input step of inputting a control command value for controlling the power source; a drive control step of controlling the power source based on the control command value to drive the vehicle; a prediction step of inputting the control command value inputted in the input step into a prediction model and predicting the driving parameter using the prediction model; and a correction step of comparing the driving parameter estimated in the measurement step with the driving parameter predicted in the prediction step, feeding back the difference between them so that the estimated driving parameter approaches the predicted driving parameter, thereby correcting the control command value inputted in the input step and inputting it into the drive control step.

[0009] According to the above means, the control command value can be corrected so that the actual driving parameters estimated in the measurement step approach the predicted driving parameters, thereby suppressing the influence of changes in the vehicle driving state.

[0010] A third means for solving the above problem is a vehicle control method implemented by a vehicle control device that controls a power source of a vehicle, and includes: a measurement step of estimating at least one of driving parameters correlated with the driving state of the vehicle based on measurement results from on-board sensors; an input step of inputting a control command value for controlling the power source; a drive control step of controlling the power source based on the control command value to drive the vehicle; a prediction step of inputting the control command value inputted in the input step into a prediction model and predicting the driving parameter using the prediction model; and a correction step (65) of comparing the driving parameter estimated in the measurement step with the driving parameter predicted in the prediction step, feeding back the difference between them so that the estimated driving parameter approaches the predicted driving parameter, thereby correcting the control command value inputted in the input step and inputting it into the drive control step.

[0011] According to the above means, the control command value can be corrected so that the actual driving parameters estimated in the measurement step approach the predicted driving parameters, thereby suppressing the influence of changes in the vehicle driving state. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. [Figure 2] FIG. 1 is a block diagram showing a vehicle control system. [Figure 3] 1 is a diagram illustrating the forces acting on a tire. [Figure 4] FIG. 2 is a diagram illustrating the relationship between sideslip angle and grip force. [Figure 5] FIG. 10 is a diagram illustrating the difference in running trajectory due to wear. [Figure 6]FIG. 2 is a block diagram showing the functions of a motor controller. [Figure 7] 10 is a flowchart showing a learning process. [Figure 8] 10 is a flowchart showing a correction process. [Figure 9] 10 is a flowchart showing a learning process according to a second embodiment. [Figure 10] FIG. 10 is a block diagram showing the functions of a motor controller according to a third embodiment. [Figure 11] 10 is a flowchart showing a learning process according to a third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, embodiments of a vehicle control device, a vehicle control program, and a vehicle control method according to the present disclosure will be described in detail with reference to the drawings. Note that, between the embodiments and modifications, the same or corresponding parts in the drawings are designated by the same reference numerals, and their descriptions will not be repeated in principle.

[0014] (First embodiment) FIG. 1 shows a vehicle 100 according to this embodiment. The vehicle 100 is assumed to be an ultra-compact mobility vehicle that automatically drives itself or a self-propelled autonomous robot (such as an automatic delivery robot). The dimensions of the vehicle 100 are assumed to be a total length of 120 to 190 cm, a width of 60 to 70 cm, and a height of 120 cm or less, with a maximum speed of approximately 6 to 20 km / h. Note that although the vehicle 100 shown in FIG. 1 has six tires 101, it is actually a two-wheel differential, and changes course using the difference in rotation between left and right drive wheels 101a, 101b of the six tires 101. Note that the vehicle may also be equipped with steering wheels.

[0015] The vehicle 100 includes a vehicle control system 10 shown in Fig. 2. The vehicle control system 10 includes a battery 20, a power control unit 30, motors 40a and 40b, a main controller 50, and a motor controller 60 as a vehicle control device.

[0016] The battery 20 is, for example, a battery pack including a series connection of unit cells, and is a DC power supply in this embodiment. The unit cell is a single battery cell or a series connection of multiple battery cells. The battery cell is, for example, a secondary battery such as a lithium-ion battery. The battery 20 converts power through the power control unit 30 and supplies the power to the motors 40a, 40b.

[0017] The power control unit 30 converts power from the battery 20 to the motors 40a, 40b in accordance with a control signal from the motor controller 60. The power control unit 30 includes, for example, an inverter that converts a DC voltage from the battery 20 into an AC voltage to drive the motors 40a, 40b, and a converter that boosts the DC voltage supplied to the inverter to a voltage equal to or higher than the output voltage of the battery 20.

[0018] Motors 40a, 40b are AC rotating electric machines, such as three-phase AC synchronous motors with permanent magnets embedded in the rotor. Motors 40a, 40b are provided for drive wheels 101a, 101b, respectively. Motors 40a, 40b are driven by power control unit 30 to generate rotational driving force, which is transmitted to drive wheels 101a, 101b, respectively.

[0019] The main controller 50 includes a CPU, ROM, RAM, input / output ports for inputting and outputting various signals, and the like. The CPU loads a program stored in the ROM into the RAM and executes it. The program stored in the ROM describes processes related to driving control of the vehicle 100. For example, the main controller 50 determines the traveling direction and speed of the vehicle 100 based on a set or stored traveling route and results from various sensors, and outputs various command values ​​based on the determined traveling direction and speed. At this time, the main controller 50 outputs control command values ​​for controlling the motors 40a and 40b to the motor controller 60. Examples of the control command values ​​include a speed command value V* of the vehicle 100 (axle center), an angular velocity command value ω*, and a turning radius r. The turning radius r may be calculated from the speed command value V* or the angular velocity command value ω*. Furthermore, when the speed of the vehicle 100 is to be kept constant, the speed command value V* does not need to be output. In this case, a command value instructing forward or backward movement may be output instead of the speed command value V*.

[0020] The motor controller 60 is primarily composed of a microcomputer including a processing unit 60a such as a CPU and a storage unit 60b such as various types of memory. The functions provided by the microcomputer can be provided by software stored in a physical memory device and a computer executing the software, software alone, hardware alone, or a combination thereof. For example, when the microcomputer is provided by hardware electronic circuits, the functions can be provided by digital circuits including numerous logic circuits or analog circuits. For example, the processing unit 60a of the microcomputer executes programs stored in a non-transitory tangible storage medium (non-transitory tangible storage medium) that serves as the storage unit 60b. The programs include, for example, programs that realize the functions shown in FIG. 6 . Execution of the programs results in the execution of a method corresponding to the programs. The storage unit 60b is, for example, a non-volatile memory. The programs stored in the storage unit 60b can be downloaded and updated via a communication network such as the Internet, for example, via OTA (Over the Air) or other means.

[0021] For example, the motor controller 60 has a function of determining the rotation speeds of the motors 40a and 40b from the angular velocity command value ω* and the like, and performing drive control of the motors 40a and 40b.

[0022] The grip force of the tire 101 gradually decreases due to wear. When the grip force of the tire 101 decreases, the side slip angle when the vehicle 100 turns a corner increases, causing a discrepancy between the arrival point and the destination point.

[0023] This will be explained in detail with reference to Figure 3. Figure 3 is a schematic diagram of a tire 101 viewed from above. As shown in Figure 3, when cornering, a centrifugal force Fc is generated on the outside (the side opposite the turning center) in a direction perpendicular to the traveling direction Dx of the vehicle 100, and a cornering force Fy (turning centripetal force) is generated on the inside (the side closer to the turning center). The angle between the tire direction Dt and the traveling direction Dx of the vehicle 100 is the sideslip angle β. The sideslip angle β can be expressed as the following equation (1).

[0024] dβ / dt=-(K / mv)×β-ω···(1) In equation (1), the sideslip angle is "β", the weight of the vehicle 100 is "m", the speed of the vehicle 100 is "v", and the angular velocity of the vehicle 100 is "ω". "K" is a constant that correlates with the grip of the tire, and the lower the grip, the smaller the value of "K". dβ / dt is the time derivative of the sideslip angle.

[0025] The relationship between this sideslip angle β and the constant "K" that correlates with grip force is as shown in FIG. 4. As shown in FIG. 4, even if the angular velocity ω of the vehicle 100 is constant, the smaller the grip force, i.e., the smaller "K", the larger the sideslip angle β. For this reason, as shown in FIG. 5, the turning radius (d → d') of a worn tire 101 is larger than that of a new tire 101 (a tire with little wear). Note that in FIG. 5, the trajectory of the worn tire 101 is shown by a solid line, and the trajectory of the new tire 101 is shown by a dashed line.

[0026] In this way, when the tire 101 wears, the side slip angle β when the vehicle 100 turns a corner increases, causing a deviation between the arrival point and the destination point. Note that although the case where the tire 101 is worn has been described here, a deviation between the arrival point and the destination point also occurs when the road surface is slippery (when the coefficient of friction is small).

[0027] Therefore, in this embodiment, even if the sideslip angle β becomes large, the motor controller 60 is provided with the following functions to correct (compensate) the traveling trajectory of the vehicle 100 so that the vehicle can reach the destination without deviation. These functions will be explained in detail below.

[0028] 6, the motor controller 60 has a function as a measurement unit 61, a function as an input unit 62, a function as a drive control unit 63, a function as a prediction unit 64, and a function as a correction unit 65. These functions may be realized by the calculation processing device 60a executing a program stored in the storage unit 60b, or may be realized by a hardware configuration such as an electronic circuit.

[0029] The measurement unit 61 acquires (inputs) measurement results from an on-board sensor 70 that measures the vehicle's driving state, and calculates (estimates) at least one of driving parameters that correlate with the vehicle's driving state based on the measurement results. In this embodiment, the on-board sensor 70 that measures the vehicle's driving state is a sensor such as an inertial measurement unit (IMU) or an on-board camera, and the measurement unit 61 calculates (estimates) an actual sideslip angle β as a driving parameter based on these measurement results.

[0030] The input unit 62 receives at least the angular velocity command value ω* as a control command value from the main controller 50.

[0031] The prediction unit 64 receives an angular velocity command value ω* from the input unit 62, inputs the value to a prediction model that predicts the traveling parameters, and predicts the traveling parameters using the prediction model. The prediction model may be an inference model (a trained neural network) learned by deep learning or the like, or may be a mathematical model that indicates the motion of the vehicle 100 (or the tire 101) derived by experiment or the like. The prediction model in this embodiment is a mathematical model, and the parameters of the mathematical model are determined (identified) by machine learning. The prediction model is stored in the storage unit 60b, and the prediction unit 64 reads out the prediction model from the storage unit 60b and uses it. A method for determining the parameters of the prediction model (a learning method) will be described later.

[0032] When the angular velocity command value ω* is input to this prediction model, the ideal sideslip angle β' is predicted. The ideal sideslip angle β' is the sideslip angle when a new or nearly new tire 101 (a tire 101 with no wear) travels on an ideal road surface.

[0033] The corrector 65 compares the actual sideslip angle β calculated by the measuring unit 61 with the sideslip angle β' predicted by the predictor 64 and inputs the difference (response difference) Δβ. The corrector 65 then feeds back this difference to calculate a compensation value Δω for correcting the angular velocity command value ω* input to the input unit 62, and adds this compensation value to the angular velocity command value ω*. In other words, if the sideslip angle β increases due to wear or the like and the traveling trajectory bulges outward compared to the ideal traveling trajectory, the angular velocity of the vehicle 100 is increased by the amount of the increase, thereby reducing the turning radius (while the speed of the vehicle 100 is constant) and bringing the traveling trajectory closer to the ideal traveling trajectory. The corrector 65 outputs the angular velocity command value ω*+Δω after the addition as the corrected angular velocity command value ω*' to the drive control unit 63.

[0034] When the corrected angular velocity command value ω*' is input, the drive control unit 63 determines the rotation speeds of the motors 40a, 40b based on the corrected angular velocity command value ω*'. Then, the drive control unit 63 performs drive control of the motors 40a, 40b so that the motors 40a, 40b are driven at the determined rotation speeds. Specifically, the drive control unit 63 outputs a control signal to the power control unit 30 for performing drive control of the motors 40a, 40b.

[0035] When the power control unit 30 receives a control signal from the motor controller 60, it performs bidirectional power conversion between the battery 20 and the motors 40a, 40b in accordance with the control signal, and supplies AC current to each of the motors 40a, 40b. When AC current is supplied from the power control unit 30, each of the motors 40a, 40b generates a rotational driving force, which is transmitted to each of the drive wheels 101a, 101b. This causes the vehicle 100 to travel.

[0036] Next, a method for determining the parameters of the prediction model (learning method) will be described. The motor controller 60 has functions as a data collection unit 66 that collects driving data and a learning unit 67 that determines a prediction model through machine learning. These functions may be realized by the arithmetic processing device 60a executing a program stored in the storage unit 60b, or may be realized by a hardware configuration such as an electronic circuit.

[0037] The motor controller 60 performs a learning process shown in Fig. 7. The learning process is performed at a predetermined timing, for example, by operating an operation terminal of the vehicle 100. Specifically, this learning process is performed when the tires 101 are new or in a nearly new condition (not worn), such as when the vehicle 100 is manufactured, when the vehicle 100 is initially configured after shipping, or when the vehicle 100 is reconfigured after maintenance. It is also desirable to perform the learning process when the road surface on which the vehicle 100 is traveling is in an ideal condition, that is, in the environment in which the vehicle 100 will actually be used.

[0038] When the learning process is performed, the motor controller 60 first starts setting a learning period (step S101). Next, the data collection unit 66 of the motor controller 60 performs test driving to collect driving data (learning data) during this learning period (step S102). A driving pattern (driving route) for the test driving is pre-stored in the memory unit 60b. The data collection unit 66 generates an angular velocity command value ω* as a control command value based on this driving pattern, and inputs the angular velocity command value ω* as is to the drive control unit 63 via the input unit 62 (indicated by a dashed line). In other words, the angular velocity command value ω* is input without being corrected by the correction unit 65. The drive control unit 63 performs drive control of the motors 40a, 40b based on the angular velocity command value ω*, and performs a test driving of the vehicle 100.

[0039] During this learning period, i.e., during the test driving, data collection unit 66 associates the actual sideslip angle β calculated by measurement unit 61 with the angular velocity command value ω* input to input unit 62 and stores them in chronological order (step S103). That is, learning unit 67 collects the history of the measured actual sideslip angle β and the input angular velocity command value ω* as a result of the test driving, and stores them in storage unit 60b or the like.

[0040] Next, the motor controller 60 determines whether the predetermined learning period has ended (step S104). If the determination result of step S104 is negative, step S102 is executed again. On the other hand, if the determination result of step S104 is positive, the learning unit 67 determines parameters of the prediction model based on the driving data collected in step S103 (step S105). Specifically, the angular velocity command value ω* of the driving data (i.e., the angular velocity command value ω* during the test driving) is input to the prediction model before learning is completed, and the sideslip angle β' output from the prediction model is compared with the actual sideslip angle β included in the driving data. Based on the comparison result, the parameters of the prediction model are identified (determined) so that the sideslip angle β' output from the prediction model approaches the actual sideslip angle β included in the driving data. In this embodiment, the prediction model is a mathematical expression model with one or more orders, and one or more parameters included in the mathematical expression model are identified.

[0041] Then, the learning unit 67 stores the prediction model for which the parameters have been determined as a trained prediction model in the storage unit 60b (step S106), and ends the learning process.

[0042] Next, a correction process for correcting the driving trajectory in response to an increase in the sideslip angle β during normal driving will be described with reference to Fig. 8. The correction process is performed by the motor controller 60 after the start of normal driving, which is not a test driving.

[0043] First, the input unit 62 of the motor controller 60 receives the angular velocity command value ω* from the main controller 50 (step S201). Next, the prediction unit 64 of the motor controller 60 receives the angular velocity command value ω* from the input unit 62, inputs it to a trained prediction model stored in the memory unit 60b, and predicts the sideslip angle β' using the prediction model (step S202).

[0044] Furthermore, the measurement unit 61 of the motor controller 60 acquires the measurement results of the on-board sensor 70 during normal driving from the on-board sensor 70, and estimates the actual sideslip angle β from the measurement results (step S203).

[0045] The correction unit 65 of the motor controller 60 compares the actual sideslip angle β calculated by the measurement unit 61 with the sideslip angle β' predicted by the prediction unit 64, and inputs the difference (response difference) Δβ (step S204).

[0046] Then, the correction unit 65 feeds back the difference and calculates a compensation value Δω for correcting the angular velocity command value ω* input to the input unit 62 (step S205).

[0047] The correction unit 65 adds the compensation value Δω to the angular velocity command value ω*, and outputs the angular velocity command value ω*+Δω after the addition as the corrected angular velocity command value ω*' to the drive control unit 63 (step S206).

[0048] When the corrected angular velocity command value ω*' is input, the drive control unit 63 determines the rotation speeds of the motors 40a, 40b based on the corrected angular velocity command value ω*'. Then, the drive control unit 63 performs drive control of the motors 40a, 40b so that the motors 40a, 40b are driven at the determined rotation speeds (step S207).

[0049] According to the above embodiment, the following effects are achieved.

[0050] The motor controller 60 corrects the angular velocity command value ω* by feeding back the difference between the measured actual sideslip angle β and the sideslip angle β' predicted based on the prediction model so that the measured actual sideslip angle β approaches the sideslip angle β' predicted by the prediction model. Specifically, the correction unit 65 of the motor controller 60 compares the actual sideslip angle β calculated by the measurement unit 61 with the sideslip angle β' predicted by the prediction unit 64, feeds back the response difference Δβ, and calculates a compensation value Δω for correcting the angular velocity command value ω* input to the input unit 62. The correction unit 65 then controls the drive of each motor 40a, 40b based on the corrected angular velocity command value ω*'. This makes it possible to arrive at the destination without deviation even if the tires 101 wear. Therefore, even if there is a change in the driving condition of the vehicle 100, such as if the tires 101 wear or the road surface is slippery, the effects of this can be suppressed.

[0051] Furthermore, the motor controller 60 adjusts the rotation speed of each of the motors 40a, 40b to correct the traveling trajectory. In other words, the traveling route is not changed by the main controller 50. This eliminates the need for the main controller 50 to perform processing to correct the traveling route, such as recalculating the traveling route.

[0052] The motor controller 60 performs a test drive of the vehicle 100, and performs machine learning based on the driving data measured during the test drive to determine the parameters of the prediction model. In this way, since the prediction model is created based on the driving data obtained by driving the actual vehicle 100, it is possible to generate a prediction model that corresponds to the specifications and driving environment of the actual vehicle 100. Therefore, it is possible to correct deviations according to the specifications and driving environment of the actual vehicle 100, and it is possible to reduce deviations from the destination. Furthermore, there is no need to prepare a prediction model that corresponds to the specifications and driving environment of the vehicle 100 in advance, which reduces the effort required.

[0053] The test run is performed when the vehicle 100 is manufactured, or when initial settings are made after shipping, or when resetting is made after maintenance. Therefore, the test run is performed when the tire 101 is in a new or nearly new condition, and appropriate running data can be collected.

[0054] (Second embodiment) A second embodiment in which the motor controller 60 of the first embodiment is partially modified will be described.

[0055] In the second embodiment, when machine learning is performed, multiple prediction models are prepared in advance, and as a result of machine learning, a model that has the highest accuracy in prediction results among the multiple prediction models is determined as the trained prediction model. Hereinafter, a method for determining a prediction model (learning method) in the second embodiment will be described with reference to FIG.

[0056] As in the first embodiment, the motor controller 60 performs the learning process shown in Fig. 9 at a predetermined timing. When the learning process is performed, the motor controller 60 first starts setting a learning period (step S301).

[0057] Next, during this learning period, the data collection unit 66 performs a test drive (step S302) in the same manner as in step S102 of the first embodiment. Also, during this learning period, the data collection unit 66 collects driving data (step S303) in the same manner as in step S103 of the first embodiment.

[0058] Next, the motor controller 60 determines whether or not a predetermined learning period has ended (step S304). If the result of this determination is negative, step S302 is executed again.

[0059] On the other hand, if the determination result in step S304 is positive, the learning unit 67 determines parameters of multiple types of prediction models based on the driving data collected in step S303 (step S305). In the second embodiment, multiple mathematical formula models with different orders are prepared as prediction models before learning. In step S305, the learning unit 67 determines parameters for each prediction model based on the driving data collected in step S303. Note that the method of determining the parameters is similar to step S105 in the first embodiment, and therefore detailed description thereof will be omitted.

[0060] Next, the learning unit 67 evaluates each prediction model determined in step S305 using the driving data collected in step S303 as test data (step S306). Specifically, the learning unit 67 inputs the angular velocity command value ω* of the test data to each prediction model determined in step S305 and obtains the resulting sideslip angle β'. The learning unit 67 then compares this sideslip angle β' with the sideslip angle β of the test data and evaluates each prediction model based on the comparison results.

[0061] Specifically, the learning unit 67 calculates an evaluation index of the prediction performance of each prediction model, compares the evaluation indexes, and identifies the prediction model with the highest evaluation. The evaluation index of the prediction performance is arbitrary, but may be, for example, NRMSE (normalized root-mean-square error).

[0062] Then, the learning unit 67 stores the prediction model with the highest evaluation as a trained prediction model in the storage unit 60b (step S307), and ends the learning process.

[0063] According to the second embodiment, the following effects are achieved.

[0064] Learning unit 67 evaluates each prediction model based on the results of a comparison between sideslip angle β', which is a predicted value output from each prediction model whose parameters have been determined as a result of machine learning, and sideslip angle β, which is an ideal value included in the driving data, and stores the prediction model evaluated as having the highest prediction accuracy as the trained prediction model. This makes it possible to generate a more appropriate prediction model according to the specifications and driving environment of vehicle 100, thereby further improving the prediction accuracy of sideslip angle β.

[0065] In this embodiment, multiple mathematical models with different orders are prepared as pre-learning prediction models, which allows for the generation of a more appropriate prediction model according to the specifications and driving environment of vehicle 100, thereby further improving the prediction accuracy of sideslip angle β.

[0066] (Third embodiment) A third embodiment in which the motor controller 60 of the first embodiment is partially modified will be described.

[0067] In the first and second embodiments, the motor controller 60 performs machine learning, but in the third embodiment, the motor controller 60 communicates with the external device 200 via a communication network 201, and causes the external device 200 to perform machine learning. The third embodiment will be described in detail below.

[0068] 10, the motor controller 60 has a function as a data collection unit 71 and a function as a communication unit 72. As in the first embodiment, the motor controller 60 has a function as a measurement unit 61, a function as an input unit 62, a function as a drive control unit 63, a function as a prediction unit 64, and a function as a correction unit 65.

[0069] The data collection unit 71 has the function of carrying out test driving during a predetermined learning period, collecting driving data, and storing the data in the storage unit 60b, similar to the first embodiment.

[0070] The communication unit 72 has a function of transmitting driving data to the external device 200 via the communication network 201, causing the external device 200 to determine a prediction model based on the driving data, and receiving the determined prediction model.

[0071] Next, a model determination process executed to determine a prediction model in the third embodiment will be described with reference to Fig. 11. The model determination process is performed by the motor controller 60. The model determination process is performed at a predetermined timing, for example, by operating an operation terminal of the vehicle 100. The timing at which the model determination process is performed is the same as the timing at which the learning process is performed in the first embodiment.

[0072] When the model determination process is performed, the motor controller 60 first starts setting a learning period (step S401). Next, during this learning period, the data collection unit 71 performs test driving to collect driving data, similar to step S102 (step S402). Furthermore, during this learning period, the data collection unit 71 collects driving data and stores it in the storage unit 60b, etc., similar to step S103 (step S403).

[0073] Next, the motor controller 60 determines whether the predetermined learning period has ended (step S404). If the determination result is negative, step S402 is executed again. On the other hand, if the determination result of step S404 is positive, the communication unit 72 of the motor controller 60 transmits the traveling data collected in step S403 to the external device 200 (step S405).

[0074] When the external device 200 receives the traveling data, it determines parameters of a prediction model based on the traveling data. Note that the method for determining the parameters of the prediction model is the same as in the first or second embodiment, and therefore detailed description thereof will be omitted. When the external device 200 determines the prediction model, it returns the prediction model (or the parameters of the prediction model) for which the pattern has been determined to the communication unit 72.

[0075] When the communication unit 72 receives the prediction model (or the parameters of the prediction model) whose parameters have been determined from the external device 200 (step S406), it stores it in the memory unit 60b as a learned prediction model (step S407) and terminates the model determination process.

[0076] The third embodiment provides the following effects.

[0077] The motor controller 60 transmits the driving data to the external device 200 via the communication network 201, and causes the external device 200 to determine the parameters of the prediction model. As a result, the motor controller 60 does not need to perform machine learning to determine the parameters of the prediction model, thereby reducing the processing load. Furthermore, the motor controller 60 does not need to have the processing performance to determine the parameters, and the required specifications for the motor controller 60 can be reduced.

[0078] (Variation) Below, modifications in which the configuration of each embodiment is partially changed will be described.

[0079] In the above embodiment, if the response difference Δβ between the sideslip angle β' (predicted value) output from the prediction model and the actual sideslip angle β (measured value) calculated by the measurement unit 61 is equal to or greater than a threshold value, the motor controller 60 may issue a warning to that effect. In other words, if the wear of the tire 101 is significant and exceeds the maximum value of the compensation value Δω (≈ if the response difference is equal to or greater than the threshold value), the motor controller 60 may issue a warning to that effect and prompt the user to perform maintenance on the tire 101.

[0080] In the above embodiment, the road surface condition may be used as a parameter to be input to the prediction model. To explain in more detail, the motor controller 60 may be provided with a road surface determination unit that acquires image information (information based on camera images) of the road surface on which the vehicle 100 is traveling and determines the road surface condition from the image information, and the prediction unit 64 may input the road surface condition together with the angular velocity command value ω* to the prediction model to predict the sideslip angle β.

[0081] In the above embodiment, the drive control unit 63 determines the rotational speed of each of the motors 40a, 40b based on the corrected angular velocity command value ω*'. However, the drive control unit 63 may determine values ​​of parameters other than the rotational speed, such as angular velocity, angular acceleration, and torque.

[0082] The control command value in the above embodiment may be a command value other than the angular velocity command value ω*'. For example, the control command value may be any one of the velocity command value V*, the angular velocity command value ω*', the turning radius r, the torque command values ​​of the motors 40a and 40b, and the rotational velocity command values ​​of the motors 40a and 40b, or a combination thereof.

[0083] In the above embodiment, a parameter other than the sideslip angle β may be used as a driving parameter correlated with the vehicle's driving state. For example, the parameters may be any one of the sideslip angle β, sideslip angular velocity, acceleration, and yaw rate, or a combination of these.

[0084] In the above embodiment, the driving pattern for the test drive was prepared (stored) in advance, but the driving pattern may be set by an operator. For example, a driving pattern corresponding to the actual driving route of the vehicle 100 may be set. This makes it possible to generate a prediction model that is more suited to the driving environment.

[0085] In the above embodiment, the driving pattern for the test drive is not limited to one pattern, and multiple patterns may be prepared.

[0086] The controller and methods described herein may be implemented by a special-purpose computer configured with a processor and memory programmed to perform one or more functions embodied in a computer program. Alternatively, the controller and methods described herein may be implemented by a special-purpose computer configured with a processor configured with one or more dedicated hardware logic circuits. Alternatively, the controller and methods described herein may be implemented by one or more special-purpose computers configured with a processor and memory programmed to perform one or more functions in combination with a processor configured with one or more hardware logic circuits. Furthermore, the computer program may be stored as instructions executed by a computer on a computer-readable non-transitory storage medium.

[0087] The following describes technical ideas that can be derived from the above-described embodiment and modifications.

[0088] [Configuration 1] A vehicle control device (10) for controlling a power source (40a, 40b) of a vehicle (100), a measurement unit (61) that estimates at least one of driving parameters correlated with the driving state of the vehicle based on a measurement result of an on-board sensor (70); an input unit (62) for inputting a control command value for controlling the power source; a drive control unit (63) that controls a power source based on a control command value to drive the vehicle; a prediction unit (64) that inputs the control command value input to the input unit into a prediction model and predicts the traveling parameters using the prediction model; a correction unit (65) that compares the driving parameters estimated by the measurement unit with the driving parameters predicted by the prediction unit, feeds back the difference between the estimated driving parameters and the predicted driving parameters, corrects the control command value input to the input unit, and inputs the control command value to the drive control unit.

[0089] [Configuration 2] a data collection unit (66) that stores the travel parameters estimated by the measurement unit as travel data while the vehicle is undergoing a test run; 2. The vehicle control device according to claim 1, further comprising a learning unit (67) that performs machine learning based on the traveling data acquired by the data collection unit to determine the prediction model.

[0090] [Configuration 3] The vehicle control device according to configuration 2, wherein the learning unit determines a plurality of types of prediction models, evaluates each prediction model based on prediction results of the plurality of prediction models, and determines the prediction model evaluated as having the highest prediction accuracy as the trained prediction model.

[0091] [Configuration 4] 4. The vehicle control device according to configuration 3, wherein the plurality of prediction models are mathematical models of different orders.

[0092] [Configuration 5] a data collection unit that stores the travel parameters estimated by the measurement unit as travel data while the vehicle is undergoing a test run; a communication unit that transmits the traveling data acquired by the data collection unit to an external device via a communication network, causes the external device to determine the prediction model based on the traveling data, and receives the determined prediction model; 2. The vehicle control device according to claim 1, wherein the prediction unit predicts the driving parameters by using a prediction model received by the communication unit.

[0093] [Configuration 6] The vehicle control device according to any one of configurations 2 to 5, wherein the test drive is performed when the vehicle is manufactured, or when initial setting is performed after shipping, or when resetting is performed after maintenance.

[0094] [Configuration 7] the power source is a rotating electric machine, The driving parameter is a sideslip angle, 7. The vehicle control device according to any one of configurations 1 to 6, wherein the control command value is an angular velocity command value of the vehicle.

[0095] [Configuration 8] a road surface determination unit that acquires image information of a road surface on which the vehicle is traveling and determines a road surface condition from the image information; A vehicle control device described in any one of configurations 1 to 7, wherein the prediction unit inputs the road surface condition determined by the road surface determination unit together with the control command value input to the input unit into the prediction model to predict the driving parameters.

[0096] [Configuration 9] The vehicle control device according to any one of configurations 1 to 8, further comprising a notification unit that issues a warning when a difference between the driving parameters estimated by the measurement unit and the driving parameters predicted by the prediction unit is equal to or greater than a threshold value.

[0097] [Configuration 10] A vehicle control program executed by a vehicle control device (10) that controls a power source (40a, 40b) of a vehicle (100), The vehicle control device a measuring step (61) of estimating at least one of driving parameters correlated with the driving state of the vehicle based on a measurement result of an on-board sensor (70); an input step (62) of inputting a control command value for controlling the power source; a drive control step (63) of controlling a power source based on a control command value to run the vehicle; a prediction step (64) of inputting the control command value input in the input step into a prediction model and predicting the traveling parameters using the prediction model; a correction step (65) of comparing the driving parameters estimated in the measuring step with the driving parameters predicted in the predicting step, correcting the control command value input in the input step by feeding back the difference therebetween so that the estimated driving parameters approach the predicted driving parameters, and inputting the corrected control command value to the drive control step.

[0098] [Configuration 11] A vehicle control method implemented by a vehicle control device (10) that controls a power source (40a, 40b) of a vehicle (100), comprising: a measuring step (61) of estimating at least one of driving parameters correlated with the driving state of the vehicle based on a measurement result of an on-board sensor (70); an input step (62) of inputting a control command value for controlling the power source; a drive control step (63) of controlling a power source based on a control command value to run the vehicle; a prediction step (64) of inputting the control command value input in the input step into a prediction model and predicting the traveling parameters using the prediction model; a correction step (65) of comparing the driving parameters estimated in the measuring step with the driving parameters predicted in the predicting step, and correcting the control command value input in the input step by feeding back the difference between the estimated driving parameters and the predicted driving parameters so that the estimated driving parameters approach the predicted driving parameters, and inputting the corrected control command value to the drive control step. [Explanation of symbols]

[0099] 40a, 40b...motor, 60...motor controller, 61...measurement unit, 62...input unit, 63...drive control unit, 64...prediction unit, 65...correction unit, 66, 71...data collection unit, 67...learning unit, 70...on-board sensor, 72...communication unit, 100...vehicle.

Claims

1. A vehicle control device (60) for controlling a power source (40a, 40b) of a vehicle (100), a measurement unit (61) that estimates at least one of driving parameters correlated with the driving state of the vehicle based on the measurement results of an on-board sensor (70); an input unit (62) for inputting a control command value for controlling the power source; a drive control unit (63) that controls a power source based on a control command value to run the vehicle; a prediction unit (64) that inputs the control command value input to the input unit into a prediction model and predicts the traveling parameters using the prediction model; a correction unit (65) that compares the driving parameters estimated by the measurement unit with the driving parameters predicted by the prediction unit, feeds back the difference between the estimated driving parameters and the predicted driving parameters, corrects the control command value input to the input unit, and inputs the control command value to the drive control unit.

2. a data collection unit (66) that stores the travel parameters estimated by the measurement unit as travel data while the vehicle is undergoing a test run; The vehicle control device according to claim 1 , further comprising a learning unit (67) that determines the prediction model by performing machine learning based on the driving data acquired by the data collection unit.

3. 3. The vehicle control device according to claim 2, wherein the learning unit determines a plurality of types of prediction models, evaluates each prediction model based on prediction results of the plurality of prediction models, and determines the prediction model evaluated as having the highest prediction accuracy as the trained prediction model.

4. The vehicle control device according to claim 3 , wherein the plurality of prediction models are mathematical models of different orders.

5. a data collection unit (71) that stores the travel parameters estimated by the measurement unit as travel data while the vehicle is undergoing a test run; a communication unit (72) that transmits the traveling data acquired by the data collection unit to an external device via a communication network, causes the external device to determine the prediction model based on the traveling data, and receives the determined prediction model; The vehicle control device according to claim 1 , wherein the prediction unit predicts the driving parameters by using a prediction model received by the communication unit.

6. The test run is performed during the manufacture of the vehicle, or during initial setup performed after shipment, or during re-setting set after maintenance. A vehicle control device according to any one of claims 2 to 5.

7. the power source is a rotating electric machine (40a, 40b), The driving parameter is a sideslip angle (β), The vehicle control device according to any one of claims 1 to 5, wherein the control command value is an angular velocity command value (ω*) of the vehicle.

8. a road surface determination unit that acquires image information of a road surface on which the vehicle is traveling and determines a road surface condition from the image information; A vehicle control device according to any one of claims 1 to 5, wherein the prediction unit inputs the road surface condition determined by the road surface determination unit, together with the control command value input to the input unit, into the prediction model to predict the driving parameters.

9. The vehicle control device according to any one of claims 1 to 5, further comprising a notification unit that issues a warning when a difference between the driving parameter estimated by the measurement unit and the driving parameter predicted by the prediction unit is equal to or greater than a threshold value.

10. A vehicle control program executed by a vehicle control device (60) that controls a power source (40a, 40b) of a vehicle (100), The vehicle control device a measurement step (61) of estimating at least one of driving parameters correlated with the driving state of the vehicle based on a measurement result of an on-board sensor (70); an input step (62) of inputting a control command value for controlling the power source; a drive control step (63) of controlling a power source based on a control command value to run the vehicle; a prediction step (64) of inputting the control command value input in the input step into a prediction model and predicting the driving parameters using the prediction model; and a correction step (65) of comparing the driving parameters estimated in the measurement step with the driving parameters predicted in the prediction step, correcting the control command value input in the input step by feeding back the difference therebetween so that the estimated driving parameters approach the predicted driving parameters, and inputting the corrected control command value to the drive control step.

11. A vehicle control method implemented by a vehicle control device (60) that controls a power source (40a, 40b) of a vehicle (100), comprising: a measurement step (61) of estimating at least one of driving parameters correlated with the driving state of the vehicle based on a measurement result of an on-board sensor (70); an input step (62) of inputting a control command value for controlling the power source; a drive control step (63) of controlling a power source based on a control command value to run the vehicle; a prediction step (64) of inputting the control command value input in the input step into a prediction model and predicting the driving parameters using the prediction model; a correction step (65) of comparing the driving parameters estimated in the measurement step with the driving parameters predicted in the prediction step, and correcting the control command value input in the input step by feeding back the difference between the estimated driving parameters and the predicted driving parameters so that the estimated driving parameters approach the predicted driving parameters, and inputting the corrected control command value to the drive control step.

Images