Control device for autonomous driving vehicle

Abstract

A control device for an autonomous vehicle according to the present invention is configured in such a manner that a control input for causing the vehicle to travel in an autonomous manner along a target trajectory obtained on the basis of an estimated current position of the vehicle is obtained by repeating a predetermined calculation, and a controller controls the vehicle to travel in an autonomous manner on the basis of the relative position of the vehicle on a travel path. The present invention estimates the current position by correcting the position of a vehicle on map data, and if it is determined that the estimation of the current position is incorrect on the basis of the amount of correction of the position at that time, adjusts a prescribed parameter in a prescribed calculation so as to reduce the number of calculations in accordance with the amount of correction.

Description

Technical Field

[0001] The present invention relates to a control device for a vehicle capable of automatically driving based on the behavior of the vehicle recognized by sensors and the like and information acquired from the outside. Background Art

[0002] Japanese Patent Application Laid-Open No. 2019-207190 discloses a vehicle position estimation device designed to stabilize the estimated accuracy of the vehicle's own position even when traveling in sections with varying lane numbers. The device estimates the vehicle's position based on image data acquired by an onboard camera, data on the vehicle's state acquired from sensors, GPS data, and map data. In the device disclosed in Japanese Patent Application Laid-Open No. 2019-207190, when a section with an increase or decrease in the number of lanes, i.e., a laneless section, is identified based on image data, the use of map data is limited compared to when no laneless sections are identified. Specifically, the weighting of the map data is reduced to correct the vehicle's estimated position. Furthermore, the device is configured to provide assistance for autonomous driving based on the vehicle's estimated position. This configuration in Japanese Patent Application Laid-Open No. 2019-207190 prevents a decrease in the estimated accuracy of the vehicle's own position in laneless sections, such as those where road segment data is inaccurately generated. Therefore, the accuracy of self-position estimation can be stabilized.

[0003] To determine the vehicle's position, components are determined for the longitudinal (front-to-back), lateral (left-to-right), and yaw (or azimuth) directions. Specifically, for each component, as in the apparatus disclosed in Japanese Patent Application Laid-Open No. 2019-207190, the vehicle's position, estimated based on vehicle state quantities and map data, and the relative position of the vehicle on the actual travel path measured by a camera, are obtained at predetermined intervals and matched to estimate the vehicle's position on the map data. As in Japanese Patent Application Laid-Open No. 2019-207190, in the case of a vehicle capable of autonomous driving, the vehicle's driving mechanisms, such as a drive unit, are controlled so that the vehicle travels along a target trajectory from this estimated position.

[0004] During such autonomous driving, calculations are periodically repeated based on data such as the vehicle's movements in various directions, map data, and estimated position, acquired as described above. By repeating such calculations, the vehicle may be controlled to follow a target trajectory within the lane it is currently traveling. In such control, from the perspective of vehicle safety, if the actual vehicle position deviates from the target trajectory by more than a specified amount, constraints may be imposed, such as altering the control inputs to the vehicle's driving mechanisms, to quickly return the vehicle to the target position. For example, the calculation equation may include a term that adds a calculation when the vehicle's position deviates by more than a specified amount.

[0005] On the other hand, when estimating the vehicle's position as described above, the estimated vehicle position may deviate significantly due to errors in the map data, unequal weighting of parameters used in estimating the position, and other factors. Consequently, while the vehicle's actual position may deviate from the target position on the target trajectory, this deviation is not immediately reflected in the control input, leading to a larger deviation. In such cases, the estimated vehicle position may need to be significantly revised, and the vehicle's behavior on the map data based on the revised estimated position may become unnatural compared to the actual vehicle's behavior. As a result, the number of repetitive calculations increases, potentially increasing the load on the control device. Furthermore, if such deviations cause a violation of the aforementioned constraints, the calculations may not be performed correctly. Furthermore, the increased number of calculations may further increase the load. Alternatively, if a constraint is suddenly violated, the vehicle's behavior may become unnatural to the passengers due to the control system's attempt to quickly return the vehicle to the target trajectory. Summary of the Invention

[0006] The present invention has been completed with an eye on the above-mentioned technical problems, and its object is to provide a control device for an autonomous driving vehicle that can reduce the load of the calculation when assisting the vehicle's driving based on control input obtained by repeatedly executing a predetermined calculation.

[0007] To achieve the above-mentioned objectives, the control device for an autonomous driving vehicle of the present invention comprises:

[0008] a plurality of behavior detection sensors that detect parameters related to the behavior of the vehicle; and

[0009] a plurality of external detection sensors that detect data related to objects existing around the road on which the vehicle is traveling,

[0010] The control device of the autonomous vehicle is configured to estimate the current position of the vehicle on the map data based on pre-stored map data, detection data from the plurality of behavior detection sensors, and detection data from the plurality of external detection sensors, determine a target trajectory based on the estimated current position of the vehicle on the map data, determine a control input for causing the vehicle to autonomously travel along the target trajectory by repeatedly performing predetermined calculations, and execute driving control of the vehicle based on the control input.

[0011] The control device of the autonomous vehicle includes a controller that estimates the current position of the vehicle and executes the driving control for causing the vehicle to travel along the target trajectory.

[0012] The controller corrects the position of the vehicle on the map data based on the relative position of the vehicle on the travel route determined from the detection data of the plurality of external detection sensors, thereby estimating the current position of the vehicle on the map data, and

[0013] calculating a correction amount for the position of the vehicle on the map data when estimating the current position of the vehicle on the map data,

[0014] Based on the correction amount, determining whether the estimation of the current position of the vehicle on the map data is correct,

[0015] When it is determined that the estimation of the current position of the vehicle on the map data is incorrect, a predetermined parameter in the predetermined calculation is adjusted according to the correction amount so as to reduce the number of calculations.

[0016] Furthermore, in the present invention, the determination of whether the estimation of the current position of the vehicle on the map data is correct may include determining whether the correction amount is greater than a prescribed correction amount that can be determined as a possible behavior of the vehicle.

[0017] Furthermore, in the present invention, the controller may include a prediction model for predicting the behavior of the vehicle until a predetermined time in the future.

[0018] The predetermined calculation in the driving control is subject to a constraint that, if the deviation between the estimated current position of the vehicle and the predicted position of the vehicle until the predetermined future time obtained based on the prediction model and the target position on the target trajectory is greater than a predetermined deviation, the control input is changed so as to reduce the deviation.

[0019] The constraint is configured such that when the deviation is larger than the predetermined deviation in the predetermined calculation, the control input is changed by performing an additional calculation.

[0020] Adjustment of the above parameters includes relaxing the above constraints.

[0021] Furthermore, in the present invention, the predetermined calculation in the driving control may be repeatedly performed until a predetermined convergence determination threshold value for determining that the control input is an appropriate value is satisfied.

[0022] The adjustment of the parameters is configured to include increasing the convergence determination threshold.

[0023] According to the control device for an autonomously driven vehicle of the present invention, the vehicle's current position is estimated by correcting the vehicle's position on map data and the vehicle's behavior detection sensors based on the vehicle's relative position, such as the relative distance between the vehicle and a white line, based on detection results from the vehicle's external detection sensors. Furthermore, a target trajectory is set based on the estimated current position of the vehicle, and driving control is performed so that the vehicle automatically drives along the target trajectory. Driving control is performed based on the amount of operation of a mechanism for driving the vehicle, determined based on control inputs calculated by repeatedly performing predetermined calculations. Furthermore, during driving control, the accuracy of the estimated current position of the vehicle on the map data is determined based on the amount of correction used in estimating the current position. If the estimated current position is determined to be incorrect, predetermined parameters in the predetermined calculation are adjusted based on the correction amount to reduce the number of calculations. Specifically, if the number of calculations in a predetermined calculation increases due to an incorrect estimate of the vehicle's current position on the map data, the predetermined calculation is adjusted based on the correction amount to reduce the number of calculations. This reduces the load on the control device caused by the increased number of calculations.

[0024] Furthermore, determining whether the estimate of the vehicle's current position on the map data is correct includes determining whether the correction amount is greater than a predetermined correction amount that indicates a possible vehicle behavior. Furthermore, this predetermined parameter is adjusted by relaxing constraints imposed on the predetermined calculation to minimize vehicle position deviations, or by relaxing a predetermined convergence threshold set to calculate appropriate control inputs for the predetermined calculation. For example, if the predetermined conditions are met, the system may perform actions such as (i) making it less likely to violate the constraints, (ii) changing the predetermined calculation to a problem that does not include the constraints, (iii) temporarily switching the control to a rule-based driving control method such as PID control, or (iv) increasing the convergence threshold. Therefore, the present invention can suppress the increase in the number of calculations caused by constraint violations and the increase in the number of calculations caused by a low convergence threshold, which in turn increases the load on the ECU. Furthermore, the present invention can prevent the failure to obtain a solution to the predetermined optimal control problem. Furthermore, it can also prevent the vehicle from exhibiting undesirable behavior.

[0025] Features, advantages, and technical and industrial significance of embodiments of the present invention will be described below with reference to the accompanying drawings, wherein like reference numerals represent like elements. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a diagram showing an example of a vehicle equipped with a control device for an autonomously driven vehicle according to an embodiment of the present invention.

[0027] Figure 2 This is a block diagram for explaining the functional structure of a control device for an autonomous driving vehicle in an embodiment of the present invention.

[0028] Figure 3 This is a flowchart showing an example of control executed by the control device of the autonomous driving vehicle in the embodiment of the present invention. DETAILED DESCRIPTION

[0029] Hereinafter, the present invention will be described based on the embodiment shown in the drawings. Note that the embodiment described below is merely an example of a case where the present invention is embodied, and does not limit the present invention.

[0030] Figure 1An autonomous vehicle (hereinafter referred to simply as a vehicle) 1 equipped with a control device according to an embodiment of the present invention is shown. Vehicle 1 according to this embodiment of the present invention is capable of not only manual driving but also automated driving by controlling driving force, braking force, steering amount, and the like without driver input. Furthermore, vehicle 1 is configured to recognize the driving environment and monitor surrounding conditions based on externally acquired peripheral information.

[0031] Vehicle 1 performs automated driving based on data related to the behavior of vehicle 1 and data obtained from the outside. For example, vehicle 1 controls vehicle 1 by calculating operating variables such as driving force, braking force, and steering control, thereby performing automated driving. During this automated driving, data related to the estimated or measured position of vehicle 1 is obtained, and vehicle 1 is able to travel along a target trajectory predetermined based on data related to the position of vehicle 1, map data, and the like. Vehicle 1 in the embodiment of the present invention is configured to accurately estimate the current position of vehicle 1, which serves as a reference for determining the operating variables of various devices used to change the behavior of vehicle 1, so that vehicle 1 performs automated driving along the target trajectory. Furthermore, vehicle 1 may be any vehicle capable of automated driving, and may be, for example, a conventional, conventional vehicle such as an engine vehicle, a hydrogen vehicle, a hybrid electric vehicle, or a fuel cell electric vehicle.

[0032] exist Figure 1 An example of such a vehicle 1 is shown in FIG. Figure 1 The illustrated vehicle 1 includes a driving force source 2 , a braking device 3 , wheels 4 , a steering device 5 , a detection unit 6 , and an ECU (Electronic Control Unit) 7 .

[0033] The driving force source 2 outputs torque for driving the vehicle 1. The driving force source 2 may include, for example, a conventionally known engine (internal combustion engine) or a motor generator, or may include both of them.

[0034] The brake device 3 is similar to a conventionally known brake device and is provided, for example, on each of the front and rear wheels 4 of the vehicle 1. An example of the brake device 3 is a friction brake such as a disc brake, a drum brake, or a powder brake, which is configured to generate a braking force in a direction to stop the rotation of each wheel 4 by generating friction through hydraulic pressure, electromagnetic force, or the like.

[0035] The steering device 5 changes the heading angle to adjust the traveling direction of the vehicle 1. The steering device 5 is the same as a conventionally known steering device, for example, a rack and pinion electric power steering device provided with an electric assist mechanism.

[0036] The detection unit 6 is a device or apparatus for acquiring various data and information required for controlling the vehicle 1. The detection unit 6 includes an internal sensor 8 for acquiring driving-related data of the vehicle 1 itself, and an external sensor 9 for acquiring information external to the vehicle 1. The internal sensor 8 includes an acceleration sensor 8a for detecting the acceleration of the vehicle 1; a gyro sensor 8b for detecting changes in the posture and orientation of the vehicle 1, such as the yaw angle (azimuth angle), as angular velocity; and a vehicle speed sensor 8c for detecting the vehicle speed based on the rotational speed of the wheels 4. The external sensor 9 includes an onboard camera 9a for acquiring captured images; and a LiDAR 9b for acquiring information about an object based on data from reflected laser light. The internal sensor 8 corresponds to the behavior detection sensor in the embodiment of the present invention, and the external sensor 9 corresponds to the external detection sensor in the embodiment of the present invention.

[0037] ECU7 is equivalent to the controller in the embodiment of the present invention, and is mainly composed of a microcomputer composed of a processor (CPU), storage elements (RAM, ROM), and input and output devices (input and output interfaces). ECU7 is configured to use data input from various sensors 8, 9 provided on the vehicle 1 or external data, pre-stored data, etc., to perform calculations in accordance with a prescribed program, and output the results of the calculations as control command signals. For example, ECU7 loads the program stored in the recording medium into the working area of ​​the storage unit through the processor and executes it, thereby performing various controls realized by the execution of the program, thereby executing functions consistent with the prescribed purpose. In addition, as Figure 2 As shown, the ECU 7 includes a vehicle position estimation unit 10, a target trajectory calculation unit 11, and a tracking control unit 12. The vehicle position estimation unit 10, the target trajectory calculation unit 11, and the tracking control unit 12 are each composed of a different microcomputer.

[0038] The vehicle position estimation unit 10 estimates the current position of the vehicle 1 on the map data based on detection data such as changes in the vehicle's 1 behavior and external information. The vehicle position estimation unit 10 detects and calculates vehicle 1 state quantities, such as vehicle speed, acceleration, and yaw rate, using the various sensors 8 and 9 described above. The vehicle position estimation unit 10 determines the vehicle's 1 position data on the map data based on these detected vehicle 1 state quantities, past estimates of the vehicle's 1 position, and stored map data. Specifically, the unit estimates the location of the vehicle 1 on the current road on the map data and, if the road is curved, the degree of curvature or yaw rate of the road. Furthermore, the vehicle position estimation unit 10 obtains the vehicle's actual relative position on the road using external sensors 9, such as the onboard camera 9a and LiDAR 9b. For example, the vehicle position estimation unit 10 measures the relative position of the lane in which the vehicle 1 is traveling, the position of the vehicle 1 in the driving lane, etc. based on the relative distance between the vehicle 1 and the white line in the driving lane, the relative distance between the vehicle 1 and the surrounding vehicles 1 or signs, etc. identified by the vehicle-mounted camera 9a.

[0039] The vehicle position estimation unit 10 is configured to determine the difference between the measured relative position of the vehicle 1 and the estimated position of the vehicle 1 on the map data. For example, when the vehicle 1 is traveling on a curved road, the estimated position of the vehicle 1 on the map data may be along the center of the lane, while the measured relative position of the vehicle 1 may be close to a dividing line (lane centerline) to the right of the lane center. In such a case, the vehicle position estimation unit 10 is configured to determine a correction amount for the estimated position data of the vehicle 1 on the map data based on the difference between the estimated position data of the vehicle 1 on the map data and the actual measured position of the vehicle 1.

[0040] The vehicle position estimation unit 10 also calculates the difference between the estimated position data of the vehicle 1 and the measured position for each of the longitudinal (front-back), lateral (left-right), and yaw components of the vehicle 1. The vehicle position estimation unit 10 changes the weighting parameters for each component based on the characteristics of these components to appropriately estimate the position of the vehicle 1. For example, the vehicle position estimation unit 10 estimates the position of the vehicle 1 by weighting data based on detection values ​​from internal sensors 8 installed in the vehicle 1 with respect to the yaw direction. Specifically, the vehicle position estimation unit 10 calculates the yaw angle of the vehicle 1 on the map data based primarily on the vehicle 1's behavior, such as speed, acceleration, and steering control amount. Furthermore, the vehicle position estimation unit 10 estimates the position on the map data by weighting data based on the vehicle 1's position measured by external sensors 9 with respect to the lateral direction of the vehicle 1. In other words, the vehicle position estimation unit 10 estimates the position on the map data based on the vehicle 1's behavior, similarly to the yaw direction, for the lateral direction of the vehicle 1. The vehicle position estimation unit 10 is configured to estimate the position by weighting data based on the measured relative position of the vehicle 1. The vehicle position estimation unit 10 estimates the current position of the vehicle 1 on the map data by repeatedly performing the above-mentioned calculations at predetermined intervals.

[0041] The target trajectory calculation unit 11 calculates the target trajectory and target curvature of the vehicle 1 based on the current position of the vehicle 1 on the map data determined by the vehicle position estimation unit 10 and information related to the route the vehicle 1 is currently traveling or the route it is scheduled to travel, also known as the planned route. Examples of route-related information include stored map data and information about white lines, signs, and other vehicles 1 on the route detected by external sensors 9. Based on this information, the target trajectory calculation unit 11 calculates a target trajectory for the vehicle 1 on the map data that allows for comfortable autonomous driving for passengers, such as by allowing the vehicle 1 to travel at an appropriate speed in the center of the lane or by entering a curved road at a curvature consistent with the curvature of the road. Furthermore, if the current position of the vehicle 1 on the map data differs from the current target trajectory or curvature of the vehicle 1, the target trajectory calculation unit 11 calculates a trajectory that allows the vehicle 1 to quickly return to the target trajectory or curvature. These white lines, signs, and other vehicles 1 correspond to objects surrounding the vehicle 1 in the present embodiment of the present invention.

[0042] For example, the target trajectory calculation unit 11 is configured to obtain lane, dividing line, and road boundary information based on map data and calculate the target trajectory based on the target position of the vehicle 1 relative to the centerline of the travel lane. If the current position of the vehicle 1 on the map data is detected to have deviated from the target trajectory, the target trajectory calculation unit 11 calculates a travel trajectory that gradually brings the current position of the vehicle 1 on the map data toward the currently set target trajectory. Specifically, the target trajectory is calculated based on parameters such as the longitudinal, lateral, and yaw deviations from the current target trajectory (i.e., the vehicle 1's target position relative to the elapsed time from the current position on the map data), the magnitude of the control input, the travel time of the vehicle 1, the permissible deviation, and the evaluation function and constraints used in optimal control.

[0043] The tracking control unit 12 is configured to control the driving of the vehicle 1, causing the vehicle 1 to travel autonomously based on data obtained from the vehicle position estimation unit 10 and the target trajectory calculation unit 11. Specifically, the tracking control unit 12 calculates the operating variables and control inputs of various devices, such as the driving force source 2, the braking device 3, and the steering device 5, that control the behavior of the vehicle 1, including the driving force, braking force, and steering amount. The tracking control unit 12 obtains data from the vehicle position estimation unit 10 regarding corrections to the position data on the map data based on the actual position of the vehicle 1 on the travel route, for each of the longitudinal, lateral, and yaw directions. Furthermore, the tracking control unit 12 obtains the target trajectory and curvature of the vehicle 1 calculated by the target trajectory calculation unit 11. Based on this obtained data, the tracking control unit 12 calculates the operating variables and control inputs of various devices that control the behavior of the vehicle 1, required for the vehicle 1 to travel from its current position on the map data according to the target trajectory and curvature.

[0044] For example, when determining an operating variable related to the steering angle of vehicle 1 based on a target trajectory, the tracking control unit 12 calculates the target steering angle and the steering torque used to achieve this target steering angle. The target steering angle is calculated using a target curvature, target yaw angle, and target lateral deviation determined based on the shape of the target trajectory, and a control gain predetermined for these parameters. As an example, the target steering angle is calculated by multiplying the target curvature, the deviation between the target yaw angle and the actual yaw angle, the deviation between the target lateral deviation and the actual lateral deviation, and the integral of the deviation between the target lateral deviation and the actual lateral deviation by the control gain. This calculation of the target steering angle is performed at predetermined intervals, and the calculation result is sent to the ECU 7 each time to control the steering mechanism. In this way, the various devices that control the behavior of vehicle 1 are controlled based on the operating variable calculated by the tracking control unit 12.

[0045] In addition, the tracking control unit 12 includes an optimal calculation unit 12a. The optimal calculation unit 12a takes into account the characteristics (dynamic characteristics) of the vehicle 1, including the performance of the device for driving control, and calculates the optimal control input to the device based on a specified optimal control problem. In addition, the specified optimal control problem is added with constraints represented by inequalities set based on the characteristics, safety, etc. of the vehicle 1, which are determined to determine whether it is above a predetermined upper limit value or below a predetermined lower limit value. The optimal calculation unit 12a repeatedly performs calculations to find the optimal solution in the specified optimal control problem based on the evaluation function and the constraints. The optimal calculation unit 12a repeatedly performs calculations until the solution is lower than a predetermined convergence judgment threshold, that is, a specified convergence judgment value, which is used to determine whether the control input is optimal. If the solution is lower than the convergence judgment value, it is determined that the optimal solution has been found. The optimal calculation unit 12a performs such optimization calculations in each predetermined control cycle. For example, when calculating the target steering angle of the vehicle 1, it calculates the target steering angle by solving a mathematical formula based on the deviations in the lateral and yaw directions of the vehicle 1, the constraints of the steering angular velocity, the weighting of these parameters, and the accumulated value within a specified time.

[0046] Furthermore, this mathematical formula includes constraints corresponding to the characteristics, performance, or behavior of vehicle 1. These constraints enable highly accurate tracking control. For example, the optimal control problem for determining the target steering angle mentioned above includes a constraint regarding lateral deviation. If the lateral deviation exceeds a specified deviation, the target steering angle is determined so that the excess is minimized. Specifically, the constraints are established using a specified state equation based on the current steering angle, the driving trajectory, and the behavior of vehicle 1, and an inequality based on the amount of excess lateral deviation. The optimal calculation unit 12a is configured to repeatedly perform optimization calculations including these constraints during each control cycle to determine control inputs for the drive power source 2, brake system 3, steering system 5, and the like. Furthermore, the lateral, yaw, and longitudinal deviations of vehicle 1 are the deviations from the target position on the target trajectory, the estimated current position of vehicle 1, and the predicted position of vehicle 1 at a specified future time, derived from a prediction model that predicts the behavior of vehicle 1 at a specified future time. Specifically, this includes the deviation between the current position of vehicle 1 and the target position, as well as the deviation between the predicted position for a predetermined future time, as derived from the vehicle's prediction model, and the target position at that predicted position. Furthermore, the predetermined optimal control problem corresponds to the predetermined calculation in the embodiments of the present invention. Furthermore, the optimal control used in such driving control may be so-called model predictive control, which uses a model to predict the behavior of vehicle 1 and determine the optimal operation.

[0047] Next, an example of the control executed by the ECU 7 of the vehicle 1 constructed in this manner will be described. Figure 3 As shown, in S1, the correction amount for correcting the position data of vehicle 1 on the map data based on the actual measured position of vehicle 1 on the travel route is calculated. As described above, in S1, if the position data of vehicle 1 on the map data, estimated primarily based on the state quantity of vehicle 1, deviates from the relative position of vehicle 1 measured primarily based on external sensors 9, the position data is corrected based on the relative position to determine the current position of vehicle 1 on the map data. In S1, the correction amounts for each of the longitudinal, lateral, and yaw directions of vehicle 1 at that time are obtained at predetermined intervals.

[0048] After the correction amount for the position data of vehicle 1 is calculated, the process proceeds to S2 to determine whether there is a problem with the estimation of the current position of vehicle 1. That is, in S2, based on the correction amount for the position data calculated in S1, it is determined whether the estimation of the current position of the vehicle is correct. Specifically, in S2, based on the correction amount calculated in S1, it is determined whether the behavior of vehicle 1 in this cycle is a correction generated as the actual behavior of vehicle 1 relative to the behavior of vehicle 1 up to the immediately previous cycle. Conversely, in S2, it is determined whether the offset in the position data of vehicle 1 is an offset generated by calculations used to estimate the position of vehicle 1. Such a determination is made, for example, based on whether the correction amount in each direction calculated in S1 is greater than a specified correction amount.

[0049] When estimating the current position of vehicle 1, the parameters referenced during the solution and the weighting of these parameters vary depending on the various directions (components) of vehicle 1. For example, as described above, there are cases where the yaw position of vehicle 1 is calculated based on vehicle 1's speed, acceleration, steering amount, and other behaviors. Furthermore, in addition to such vehicle 1 behaviors, the lateral position is calculated by referring to data based on the relative position of vehicle 1 measured by external sensors 9 and the like. In such cases, if the actual position of vehicle 1 on the travel path deviates from the position of vehicle 1 on the map data due to errors or interference in the map data, there is a possibility that differences in the detection of this deviation may occur in the various directions of vehicle 1.

[0050] For example, when entering a curve, the position of vehicle 1 in the map data may be detected further back than its actual position due to errors, interference, and other factors as described above. In such cases, the lateral position of vehicle 1 is estimated based on data based on vehicle 1's behavior and the measured relative position of vehicle 1, allowing for relatively early detection of such deviations. On the other hand, the position of vehicle 1 in the yaw direction is estimated based on data based on vehicle 1's behavior, so detection of such deviations may be delayed relative to the lateral direction of vehicle 1. Specifically, since relative position in the yaw direction is not referenced or is weighted less heavily, the yaw angle in the map data is estimated to vary along the road, resulting in a delay in detecting such deviations.

[0051] The steering angle operation amount of vehicle 1 is calculated using the deviation between the target position of vehicle 1 on the map data and the estimated and predicted positions in the lateral and yaw directions of vehicle 1. Therefore, due to the small deviation in the yaw direction, the steering angle operation amount may not be large enough, and the lateral deviation of vehicle 1 may become large. In this case, the correction amount used to correct the position of vehicle 1 on the map data may become a large correction amount that would not be caused by the actual behavior of vehicle 1. If the correction amount calculated in S1 is determined to be larger than the prescribed correction amount that would be caused by the actual behavior of vehicle 1, S2 determines that the correction is caused by the estimation of the current position of vehicle 1. Thus, if it is determined in S2 that the correction is not caused by the estimation of the current position of vehicle 1, for example, because the correction amount is less than the prescribed correction amount, subsequent control is not executed and the process is temporarily terminated.

[0052] On the other hand, if the correction is determined to be due to estimation of the vehicle 1's position, for example, due to the correction amount being greater than a predetermined amount, the process proceeds to S3, where parameters in the optimal control problem are adjusted based on the correction amount. In S3, parameters in the predetermined optimal control problem that increase the number of calculations due to the correction amount being greater than the predetermined amount are adjusted. For example, in S3, constraints related to the vehicle 1's state and convergence thresholds in the predetermined optimal control problem are modified.

[0053] When changing constraints related to the state variables of vehicle 1, constraints set for deviations in the lateral, yaw, and longitudinal directions of vehicle 1 are relaxed based on the corrections calculated in S1. For example, if it is determined that the current position of vehicle 1 cannot be accurately estimated in the lateral direction, the constraints are relaxed by increasing the specified lateral deviation used to determine if the constraints are violated. Alternatively, the weight of the evaluation function related to the constraints in the optimal control problem is reduced, for example, the weight of the evaluation function related to the slack variables that constitute the constraints in the optimal control problem is reduced. Alternatively, the optimal control problem for lateral deviation is changed to an optimal control problem without constraints and optimization is performed. Alternatively, instead of using optimal control, a rule-based driving control method such as conventional PID control is used to perform driving control. Alternatively, the constraints in the optimal control problem may be relaxed using conventional minimum time methods, constraint relaxation methods, or the like. Furthermore, when relaxing the constraints, the relaxation is performed by taking into account hard constraints and soft constraints.

[0054] Furthermore, when changing the convergence threshold in a predetermined optimal control problem, the convergence threshold is relaxed based on the correction amount calculated in S1. For example, if it is determined that the current position of vehicle 1 cannot be accurately estimated in the lateral direction, the convergence threshold is increased in the optimal control problem used to determine the target steering angle. In this manner, in S3, the predetermined parameters in the optimal control problem are changed based on the correction amount, and the process is temporarily terminated.

[0055] The control device for the thus configured autonomous vehicle 1 estimates the current position of the vehicle 1 on the map data at predetermined intervals based on the vehicle 1's position data on the map data, estimated based on the vehicle's behavior, and the vehicle 1's actual position on the road, measured relatively by identifying surrounding objects. A target trajectory for the vehicle 1 is calculated based on the thus-calculated current position of the vehicle 1, the map data, the white lines of the lane detected by external sensors 9, the road shape, and the like. To ensure autonomous driving of the vehicle 1 along this target trajectory, a predetermined optimal control problem is solved that takes into account the constraints of the vehicle 1. The control system then calculates the operational variables of the drive force source 2, the brake device 3, the steering device 5, and the like, and the vehicle 1 then performs autonomous driving based on these operational variables.

[0056] In such driving control, a correction amount is calculated when estimating the current position of vehicle 1. At this time, based on the calculated correction amount, a determination is made as to whether the current position of vehicle 1 is accurately estimated. Specifically, a determination is made as to whether the current position of vehicle 1 on the map data has undergone a change that would not occur in the actual behavior of vehicle 1. If such a change is determined to have occurred, the constraints and convergence determination value are relaxed in a predetermined optimal control problem. In other words, the parameters that increase the number of calculations due to such a change are adjusted, taking into account the impact on driving control. For example, if a large correction amount is obtained that would not occur in the actual behavior of vehicle 1, the number of calculations required to calculate the optimal solution in the predetermined optimal control problem increases accordingly. Furthermore, if such a change violates a constraint, the number of calculations based on the constraint increases. Alternatively, the number of calculations required to reach the convergence determination value in the predetermined optimal control problem increases.

[0057] In the control device of the embodiment of the present invention, when such a change is detected, changes are made, such as relaxing the constraint, performing calculations on a problem that does not include the constraint itself, switching the control to rule-based control such as PID control, or increasing the convergence threshold. In other words, the parameters associated with the increase in the number of calculations caused by such a change are adjusted. This can prevent the load on the ECU 7 from increasing the number of calculations, or prevent the calculations from failing to find a solution to the predetermined optimal control problem. Furthermore, it can prevent the vehicle 1 from exhibiting behaviors that are undesirable to passengers.

[0058] While the embodiments of the present invention have been described above, the present invention is not limited to the aforementioned examples and may be appropriately modified within the scope of achieving the objectives of the present invention. For example, in the case of relaxing a constraint, the control may be executed when, based on the correction amount in the position data of vehicle 1, it is determined that there is a high probability of violating the constraint. That is, the change in the correction amount is predicted by comparing the change in the correction amount in the previous cycle with the correction amount in the current cycle. Furthermore, the control may be configured such that, if it is assumed that the constraint will be violated if the change continues, the constraint is relaxed or driving control is switched to unconstrained driving control to execute driving control.

Claims

1. A control device for an autonomous driving vehicle, wherein: have: a plurality of behavior detection sensors that detect parameters related to the behavior of the vehicle; and a plurality of external detection sensors that detect data related to objects existing around a road on which the vehicle is traveling, The control device of the autonomous driving vehicle is configured to estimate the current position of the vehicle on the map data based on pre-stored map data, detection data of the plurality of behavior detection sensors, and detection data of the plurality of external detection sensors, obtain a target track based on the estimated current position of the vehicle on the map data, obtain a control input for causing the vehicle to autonomously drive along the target track by repeatedly performing a predetermined operation, and execute driving control of the vehicle based on the control input. The control device of the autonomous driving vehicle includes a controller that estimates the current position of the vehicle and executes the travel control for causing the vehicle to travel along the target track. The controller corrects the position of the vehicle on the map data based on the relative position of the vehicle on the travel road obtained from the detection data of the plurality of external detection sensors, thereby estimating the current position of the vehicle on the map data, and calculating a correction amount of the position of the vehicle on the map data when estimating the current position of the vehicle on the map data, determining whether the estimation of the current position of the vehicle on the map data is correct based on the correction amount, When it is determined that the estimation of the current position of the vehicle on the map data is incorrect, a predetermined parameter in the predetermined calculation is adjusted according to the correction amount so as to reduce the number of calculations.

2. The control device for an autonomous driving vehicle according to claim 1, wherein: The determination of whether the estimation of the current position of the vehicle on the map data is correct includes determining whether the correction amount is larger than a predetermined correction amount that can be determined as a behavior that may occur as the behavior of the vehicle.

3. The control device for an autonomous driving vehicle according to claim 1 or 2, wherein: The controller has a prediction model for predicting the behavior of the vehicle until a predetermined time in the future. The prescribed operation in the driving control is provided with a constraint that, when a deviation between the estimated current position of the vehicle and the predicted position of the vehicle until the prescribed time in the future obtained based on the prediction model and the target position on the target track is greater than a prescribed deviation, the control input is changed in a manner to reduce the deviation. The constraint is configured such that when the deviation is larger than the predetermined deviation in the predetermined calculation, the control input is changed by adding a calculation, The adjusting of the parameters includes relaxing the constraints.

4. The control device for an autonomous driving vehicle according to claim 1 or 2, wherein: The predetermined calculation in the driving control is configured to be repeatedly performed until a predetermined convergence determination threshold for determining that the control input has an appropriate value is satisfied. The adjustment of the parameter includes increasing the convergence determination threshold.

Images