VEHICLE CONTROL UNIT, PROGRAM AND VEHICLE CONTROL PROCEDURES

Abstract

Vehicle control device (100), comprising: a sensor signal acquisition unit (111) configured to acquire sensor signals from a plurality of sensors sensing a physical quantity relating to a vehicle's environment; an image data acquisition unit (111) configured to acquire image data displaying images of the vehicle's environment from at least one camera capturing the images; a physical potential computation unit (114a) configured to use a pre-learned model for predicting potential risk to predict a potential risk based on a characteristic quantity of a target vehicle's environment, in order to compute a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;a human vision computation unit (112) configured to use a pre-calculated human vision model to calculate a visibility repulsion potential, wherein the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; a potential correction unit (114b) configured to calculate an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; a target path generation unit (115) configured to generate a target path to cause the vehicle to move from a current position of the vehicle to a target point in accordance with a gradient calculated from the integrated repulsion potential; and a vehicle control unit (116) configured to control the vehicle to travel along the target path.

Description

TECHNICAL AREA The disclosure relates to a vehicle control device, a program, and a vehicle control method. TECHNICAL BACKGROUND Revised description pages 2, 2a In recent years, technologies for autonomous vehicle driving have been developed. The fundamental techniques for driving vehicles are perception, judgment, and manipulation. Judgment is the processing of complex combinations of road users in indefinite numbers. Motion control for route planning and vehicle control is well-studied in terms of such judgment. For steering and controlling a vehicle, an artificial potential method (also called a potential field method) described in PTL 1 is frequently used. This artificial potential method is based on the idea that human perception of danger is expressed in vehicle behavior and essentially enables control using a physical vehicle control model. German patent DE 10 2013 013 747 A1 discloses a driver assistance system for a vehicle comprising a detector, a vector field generator, and a control device. Objects are detected, and based on this, an environmental potential field is created, which is then used to control the vehicle. German patent DE 10 2018 123 896 A1 discloses how, for the operation of at least a partially automated vehicle, objects in the environment are detected from environmental sensor data and an environmental model is created from which a trajectory for the vehicle is calculated. A potential field is used and gradient-based optimization is performed to determine the trajectory. STATE OF THE ART REFERENCES PATENT REFERENCE Patent Literature 1: Japanese Patent Application Publication No. JP 2018 - 192 954 A SUMMARY OF THE INVENTION TASK TO BE SOLVED BY THE INVENTION However, conventional vehicle control using artificial potential methods does not necessarily correspond to vehicle control through passenger recognition, assessment, and manipulation, making comfortable boarding impossible. Accordingly, one objective of one or more aspects of the disclosure is to make the autonomous driving of a vehicle possible in such a way as to reduce the inconvenience to a passenger. MEANS OF SOLVING THE TASK A vehicle control device according to one aspect of the disclosure comprises: a sensor signal acquisition unit configured to acquire sensor signals from a plurality of sensors that detect a physical quantity relating to a vehicle's environment; an image data acquisition unit configured to acquire image data displaying images of the vehicle's environment from at least one camera capturing the images; a physical potential calculation unit configured to use a pre-learned model for predicting potential risk to predict a potential risk based on a characteristic quantity of a target vehicle's environment, in order to calculate a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;a human vision computation unit configured to use a pre-calculated human vision model to calculate a visibility repulsion potential, wherein the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; a potential correction unit configured to calculate an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; a target path generation unit configured to generate a target path to cause the vehicle to move from a current position of the vehicle to a target point in accordance with a gradient calculated from the integrated repulsion potential; and a vehicle control unit configured to control the vehicle to travel along the target path. A program according to one aspect of the disclosure causes a computer to operate as: a sensor signal acquisition unit configured to acquire sensor signals from a plurality of sensors sensing a physical quantity relating to a vehicle's environment; an image data acquisition unit configured to acquire image data displaying images of the vehicle's environment from at least one camera capturing the images; a physical potential computation unit configured to use a pre-learned model for predicting potential risk to predict a potential risk based on a characteristic quantity of a target vehicle's environment, in order to compute a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;a human vision computation unit configured to use a pre-calculated human vision model to calculate a visibility repulsion potential, wherein the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; a potential correction unit configured to calculate an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; a target path generation unit configured to generate a target path to cause the vehicle to move from a current position of the vehicle to a target point in accordance with a gradient calculated from the integrated repulsion potential; and a vehicle control unit configured to control the vehicle to travel along the target path. A vehicle control method according to one aspect of the disclosure comprises: acquiring sensor signals from a plurality of sensors that detect a physical quantity relating to a vehicle's environment; acquiring image data displaying images of the vehicle's environment from at least one camera capturing the images; using a pre-learned model to predict a potential risk based on a characteristic quantity of a target vehicle's environment to calculate a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;Using a pre-calculated human vision model to calculate a visibility repulsion potential, where the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; calculating an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; generating a target path to cause the vehicle to move from its current position to a target point in accordance with a gradient calculated from the integrated repulsion potential; and controlling the vehicle to travel along the target path. EFFECTS OF THE INVENTION According to one or more aspects of the disclosure, autonomous driving of a vehicle can be carried out in such a way as to reduce the inconvenience to a passenger. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a block diagram showing the main configuration of a vehicle control system installed in an autonomous vehicle. Fig. 2 is a block diagram schematically representing the configuration of a human vision computation unit and a processing unit for the potential risk prediction model. Fig. 3 is a schematic representation of an example of the environment of a vehicle in which the vehicle control system is installed. Figs. 4A and 4B are block diagrams showing hardware configuration examples. MODE FOR EXECUTING THE INVENTION FIRST VERSION Fig. 1 is a block diagram showing the main configuration of a vehicle control system 100 installed in an autonomously driving vehicle. The vehicle control system 100 comprises a sensor group 101, a vehicle control unit 110, an actuator 102 and a vehicle drive unit 103. Sensor group 101 comprises several sensors that detect physical quantities relating to the vehicle's environment. Sensor group 101 comprises one or more cameras that function as an imaging unit and capture images of the vehicle's surroundings. The image data from the captured images is transmitted to the vehicle control unit 110. Here, sensor group 101 includes an environmental sensor that detects people and obstacles in the vicinity of the vehicle. Sensor group 101 includes, for example, at least one high-frequency radar sensor, an ultrasonic sensor, or a LiDAR sensor. The environmental sensor transmits sensor signals indicating the detected content to the vehicle control unit 110. Sensor group 101 comprises a vehicle sensor that detects the tamper state and behavior of the vehicle to which the vehicle control system 100 is mounted. Sensor group 101 includes, for example, at least one vehicle speed sensor, one acceleration sensor, and one angular velocity sensor. It also includes, for example, at least one accelerator pedal position sensor, one brake stroke sensor, one brake pressure sensor, one steering angle sensor, one engine speed sensor, one brake light switch, and one turn signal switch. The vehicle sensors transmit sensor signals indicating the detected data to the vehicle control unit 110. Sensor group 101 comprises a GPS receiver (Global Positioning System), which functions as a GPS receiving unit and contains a GPS antenna for receiving GPS signals. The received GPS signals are forwarded to the vehicle control unit 110. The vehicle control unit 110 functions as a vehicle control device that controls a vehicle on which the vehicle control system 100 is mounted. The vehicle control unit 110 comprises a sensor signal processing unit 111, a human vision calculation unit 112, a risk prediction knowledge database (hereinafter referred to as risk prediction knowledge DB) 113, a processing unit 114 for the model for predicting a potential risk, a target path generation unit 115 and a vehicle control unit 116. The sensor signal processing unit 111 acquires various signals and data from the sensor group 101, performs signal processing as required, such as analog-to-digital conversion, and passes the processed signals and data on to the subsequent components. The sensor signal processing unit 111, for example, functions as an image data acquisition unit, capturing image data from one or more cameras. The sensor signal processing unit 111 then forwards the image data to the human vision calculation unit 112 and the processing unit for the model 114 to predict potential risk. The sensor signal processing unit 111 also functions as a sensor signal acquisition unit, which acquires sensor signals from the vehicle sensors, the environment sensor and the GPS receiver. The sensor signal processing unit 111 forwards the sensor signals from the vehicle sensors to the processing unit for model 114 to predict a potential risk. The sensor signal processing unit 111 forwards the sensor signals from the environment sensor and the GPS signals, or sensor signals from the GPS receiver, to the target path generation unit 115. The Human Vision Computing Unit 112 uses a pre-defined mathematical model of human vision to calculate a visibility repulsion potential. This potential represents the detection of an image displayed by the image data as a repulsion potential energy influenced by human vision. The calculated visibility repulsion potential is then passed to the Model Processing Unit 114 to predict potential risk. For example, somatological or psychophysical models of human vision include a lateral inhibition model, in which the optic nerve cells inhibit the surrounding cells simultaneously with stimulation, e.g., by brightness; a motion perception model of optical flow, apparent motion, or the like; and a model of color vision; and these parts of the models are mathematically modeled. The Human Vision Computing Unit 112 calculates a psychological potential field from such a mathematical model and uses it as a visibility repulsion potential. Fig. 2 is a block diagram that schematically represents the configuration of the human vision computation unit 112 and the processing unit for the model 114 for predicting a potential risk. The Human Vision Computing Unit 112 comprises a Lateral Inhibition Model Processing Unit 112a and a Motion Perception Model Processing Unit 112b. The processing unit 112a for the lateral inhibition model uses a lateral inhibition model as a visual model to calculate a psychological potential as a visibility rejection potential. For example, the processing unit 112a for the lateral inhibition model uses the lateral inhibition model, which mathematically models human lateral inhibition, to calculate a psychological repulsion potential that affects the ability to detect an obstacle through human lateral inhibition as at least part of the visibility repulsion potential. According to the lateral inhibition model, the peripheral part inhibits the central part when the same stimulation of the same intensity acts on both the central and peripheral parts of the receptive field. If the stimulation acting on the central and peripheral parts differs in intensity, the difference is amplified by lateral inhibition, impairing contour and contrast enhancement. It is therefore assumed that lateral inhibition impairs the driver's ability to detect obstacles. Here, the model of lateral inhibition described in the following literature 1 is used as a reference from an engineering perspective. References 1: Katsuhiko Fujii, Akira Matsuoka, Tatsuya Morita, “Analysis of the Optical Illusion by Lateral Inhibition,” Japanese Journal of Medical Electronics and Biological Engineering, Vol. 5, Issue 2, pp. 25-34, 1967 The model of lateral inhibition can be represented as object extraction using equation (1): where i(ξ,η) is a stimulus figure, p(x,y) is the intensity of neuronal activity and w(ξ-x, η-y) is a coupling function. It should be noted that x, y, ξ and η are coordinate values ​​on the retina, x and y are reference coordinates of p, and ξ and η are coordinates of an obstacle. The receptive field of retinal neurons is a concentric structure and can be approximated by a Gaussian difference function (DOG), as shown in equation (2), when the coupling function is considered as a spatial property: where K1 is a coefficient indicating the strength of the excitatory coupling, K2 is a coefficient indicating the strength of the inhibitory coupling, σ1 is the variance of excitability, and σ2 is the variance of the inhibitory coupling. Based on the above explanations, a psychological potential Up is calculated using equation (3) if the output p(x,y) is the psychological potential Up for an object shape: In equation (3), i(ξ,η) is a stimulus figure and its effect can be considered an obstacle shape when treated as a higher-order color vision mechanism and as a binary image. Therefore, it can be considered a constant integral, and equation (4) is obtained as follows: Accordingly, the processing unit 112a uses equation (4) for the lateral inhibition model to calculate the psychological potential Upzu. The motion perception model processing unit 112b uses a motion perception model as a visual model to calculate a motion perception repulsion potential as a visibility repulsion potential. For example, the motion perception model processing unit 112b uses the motion perception model, which mathematically models human motion perception, to calculate the motion perception repulsion potential, which affects the ability to detect an obstacle through human motion perception as at least part of the visibility repulsion potential. Motion perception is calculated by determining the optical flux, apparent motion, induced motion, etc., using models such as an optical flux detection model for a moving object, e.g., the Lucas-Canada method, or a motion perception model such as the Reichardt type model, which is a computational approach, or a gradient detection model. In this case, if a pixel of a target object is pixel I(x,y,t), the pixel is calculated after Δt seconds using equation (5): A constraint equation for the movement speed of the pixels is calculated using equations (6) and (7) by Taylor extension of equation (5) and its division by dt: From this, the repulsion potential Um is calculated for the perception of motion according to equation (8): where m is the mass of the moving body and I is the distance traveled. The apparent motion and the induced motion can also be calculated based on the concept described above. Referring again to Fig. 1, a risk prediction knowledge database 113 stores a model for predicting a potential risk. The potential risk prediction model is a learned model that incorporates into a control model the repulsion potential of road boundaries, stationary obstacles, and the dive-behind-behind-stationary-obstacles, as well as the attraction potential of a vehicle trajectory (a path) towards a destination. It is assumed that human hazard perception manifests itself through manipulation, as described in PTL 1, etc. The potential risk prediction model can be constructed using known techniques. The processing unit 114 of the potential risk prediction model corrects the physical repulsion potential calculated using the potential risk prediction model stored in the risk prediction knowledge DB 113 by using the visibility repulsion potential calculated by the human vision calculation unit 112 to calculate an integrated repulsion potential. As shown in Fig. 2, the processing unit for the model for predicting a potential risk 114 comprises a physical potential calculation unit 114a and a potential correction unit 114b. The physical potential calculation unit 114a uses the model stored in the risk prediction knowledge DB 113 to predict a potential risk in order to calculate a physical repulsion potential, which is determined from the physical positional relationship between the vehicle on which the vehicle control system 100 is mounted and an object such as a person or an obstacle from the sensor signals of the sensor signal processing unit 111. The unit for calculating the physical potential 114a uses the pre-learned model for predicting a potential risk from characteristic quantities of the target vehicle's environment to calculate the physical repulsion potential, which is a repulsion potential caused by the environment of the vehicle on which the vehicle control system 100 is mounted, based on the physical quantities indicated by the sensor signals and the images displayed by the image data. As shown in Fig. 3, for example, it is assumed that a vehicle 150, on which the vehicle control system 100 is mounted, is traveling in the direction of arrow D on a straight road with walls on both sides and no lane markings. It is also assumed that a vehicle 151 is parked on the left side of the road relative to vehicle 150, that vehicle 150 is about to pass vehicle 151 on the right, and that there are no other road users. In such a situation, the associated repulsion potential corresponds to a road boundary and the repulsion of obstacles, the control model of vehicle 150 corresponds to lateral control, and vehicle 150 is intended to perform a linear movement at a constant speed. Due to the above-mentioned restrictions, a function of the repulsion potential Uw(x,y) is expressed by equation (9): where ywc is the y-coordinate of the road center, w is the weight coefficient and σw is the variance. A function of the repulsion potential Uo(x,y) is expressed by equation (10): where xor is the x-coordinate of the rear part of the parked vehicle 151, xof is the x-coordinate of the front part of the parked vehicle 151, yo is the y-coordinate of the center of the parked vehicle 151 in the width direction, where is the weight coefficient and σox and σoy are variance. Equation (9) is a one-dimensional Gaussian function where the length of the wall is infinite, and equation (10) is a two-dimensional Gaussian function. The potential correction unit 114b receives the repulsion potential Uwan of the road boundary calculated by the function of the repulsion potential Uw(x,y) and the repulsion potential Uoan of an obstacle calculated by the function of the repulsion potential Uo(x,y). The potential correction unit 114b corrects the physical repulsion potential calculated by the physical potential calculation unit 114a by using the visibility repulsion potential calculated by the human vision calculation unit 112 to calculate the integrated repulsion potential. For example, the potential correction unit 114b calculates an integrated repulsion potential Ua using equation (11): Since equation (4) is set up in a coordinate system on the retina, α is used here for the magnification in order to compare the psychological potential with the other terms set up in the real world. The magnification should be calculated in advance through experiments, etc. Here, the potential correction unit 114b adds the visibility repulsion potential to the physical repulsion potential to calculate the integrated repulsion potential; alternatively, the integrated repulsion potential can also be calculated by multiplying the physical repulsion potential with the visibility repulsion potential. Referring again to Fig. 1, the target path generation unit 115 generates a target path from the GPS signals of the sensor signal processing unit 111 and the integrated repulsion potential of the processing unit 114 for the model for predicting a potential risk. This target path is a driving route of the vehicle on which the vehicle control system 100 is mounted. The target path generation unit 115, for example, generates a target path from the current position, indicated by the GPS signals, to a destination point specified by the vehicle's driver, in order to guide the vehicle according to a gradient calculated from the integrated repulsion potential. The generated target path is then transmitted to the vehicle control unit 116. Target paths with a potential can be generated using known methods, such as the method described in PTL 1. Specifically, the target path generation unit 115 uses the potential field method to calculate the gradient of the potential as the force acting on the vehicle. The target path generation unit 115 determines a target yaw rate and a target velocity with a low integrated repulsion potential along the vehicle's trajectory a few seconds in the future. The vehicle control unit 116 controls the vehicle on which the vehicle control system 100 is mounted, in order to guide the vehicle along the target path from the target path generation unit 115. For example, the vehicle control unit 116 generates a control signal from the target path generation unit 115, which controls the vehicle to travel along the target path, and passes the control signal to the actuator 102. Specifically, the vehicle control unit 116 converts the target yaw rate determined by the target path generation unit 115 into a target steering angle. The vehicle control unit 116 also converts the target speed determined by the target path generation unit 115 into a torque. The vehicle control unit 116 then generates control signals that specify the resulting target steering angle and torque. The actuator 102 actuates the vehicle drive unit 103, which is a mechanism for propelling a vehicle comprising an engine, an accelerator pedal, a brake and a steering wheel, in accordance with the control signals from the vehicle control unit 116. The vehicle drive unit 103 is a mechanism for propelling a vehicle, comprising a motor, an accelerator pedal, a brake and a steering wheel. Some or all of the sensor signal processing unit 111, the human vision calculation unit 112, the processing unit for the model for predicting a potential risk 114, the target path generation unit 115, and the vehicle control unit 116 described above can, for example, be implemented as a central processing unit (CPU) by a memory 10 and a processor 11, which executes the programs stored in the memory 10, as shown in Fig. 4A. Such programs can be provided over a network or can be recorded and provided on a recording medium. That is, such programs can, for example, be provided as a program product. Some or all of the sensor signal processing unit 111, the human vision computation unit 112, the potential risk prediction model processing unit 114, the target path generation unit 115 and the vehicle control unit 116 can, for example, also be implemented by a single circuit, a composite circuit, a program-controlled processor, a program-controlled parallel processor, a processing circuit 12 such as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA), as shown in Fig. 4B. As described above, the sensor signal processing unit 111, the human vision calculation unit 112, the processing unit for the model for predicting a potential risk 114, the target path generation unit 115 and the vehicle control unit 116 can be implemented by processing circuits. The risk prediction knowledge DB 113 can be implemented using a storage medium, e.g., a hard disk drive (HDD), a solid state drive (SDD), or non-volatile memory. REFERENCE MARK LIST 100 Vehicle control system; 101 Sensor group; 102 Actuator; 103 Vehicle drive unit; 110 Vehicle control unit; 111 Sensor signal processing unit; 112 Human vision calculation unit; 112a Lateral inhibition model processing unit; 112b Motion perception model processing unit; 113 Risk prediction knowledge database; 114 Potential risk prediction model processing unit; 114a Physical potential calculation unit; 114b Potential correction unit; 115 Target path generation unit; 116 Vehicle control unit.

Claims

Vehicle control device (100), comprising: a sensor signal acquisition unit (111) configured to acquire sensor signals from a plurality of sensors sensing a physical quantity relating to a vehicle's environment; an image data acquisition unit (111) configured to acquire image data displaying images of the vehicle's environment from at least one camera capturing the images; a physical potential computation unit (114a) configured to use a pre-learned model for predicting potential risk to predict a potential risk based on a characteristic quantity of a target vehicle's environment, in order to compute a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;a human vision computation unit (112) configured to use a pre-calculated human vision model to calculate a visibility repulsion potential, wherein the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; a potential correction unit (114b) configured to calculate an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; a target path generation unit (115) configured to generate a target path to cause the vehicle to move from a current position of the vehicle to a target point in accordance with a gradient calculated from the integrated repulsion potential; and a vehicle control unit (116) configured to control the vehicle to travel along the target path. Vehicle control device (100) according to claim 1, wherein the human vision calculation unit (112) is configured to use a lateral inhibition model that mathematically models human lateral inhibition to calculate a psychological repulsion potential that influences the ability to detect an obstacle through human lateral inhibition as at least a part of the visibility repulsion potential. Vehicle control device (100) according to claim 1, wherein the human vision calculation unit (112) is configured to use a motion perception model that mathematically models human motion perception in order to calculate a motion perception repulsion potential that influences the ability to detect an obstacle by human motion perception as at least a part of the visibility repulsion potential. Vehicle control device (100) according to one of claims 1 to 3, wherein the potential correction unit (114b) is configured to add or multiply the physical repulsion potential to or with the visibility repulsion potential in order to correct the physical repulsion potential. Program that causes a computer to operate as: a sensor signal acquisition unit (111) configured to acquire sensor signals from a plurality of sensors sensing a physical quantity relating to a vehicle's environment; an image data acquisition unit (111) configured to acquire image data displaying images of the vehicle's environment from at least one camera capturing the images; a physical potential computation unit (114a) configured to use a pre-learned model for predicting potential risk to predict a potential risk based on a characteristic quantity of a target vehicle's environment, in order to compute a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;a human vision computation unit (112) configured to use a pre-calculated human vision model to calculate a visibility repulsion potential, wherein the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; a potential correction unit (114b) configured to calculate an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; a target path generation unit (115) configured to generate a target path to cause the vehicle to move from a current position of the vehicle to a target point in accordance with a gradient calculated from the integrated repulsion potential; and a vehicle control unit (116) configured to control the vehicle to travel along the target path. Vehicle control method, comprising: Acquiring sensor signals from a plurality of sensors that detect a physical quantity relating to a vehicle's environment; Acquiring image data displaying images of the vehicle's environment from at least one camera capturing the images; Using a pre-learned model to predict a potential risk, to predict a potential risk based on a characteristic quantity of a target vehicle's environment, to calculate a physical repulsion potential based on the physical quantity and the images, wherein the physical repulsion potential is a repulsion potential caused by the vehicle's environment;Using a pre-calculated human vision model to calculate a visibility repulsion potential, where the visibility repulsion potential is a repulsion potential influenced by the vision of a human recognizing the images; calculating an integrated repulsion potential obtained by correcting the physical repulsion potential with the visibility repulsion potential; generating a target path to cause the vehicle to move from its current position to a target point in accordance with a gradient calculated from the integrated repulsion potential; and controlling the vehicle to travel along the target path.

Images