Vehicle behavior control device

Abstract

This vehicle behavior control device comprises: a control target generation unit that generates a control target on the basis of a driving operation amount; a vehicle state acquisition unit that acquires a vehicle state; an environment information acquisition unit that acquires vehicle environment information; a model parameter estimation unit that estimates a model parameter relating to a vehicle model on the basis of the vehicle state and the vehicle environment information; a future behavior prediction unit that predicts a future behavior of the vehicle by applying the vehicle state and the model parameter to the vehicle model; a vehicle control amount generation unit that applies the generated control target and the predicted future behavior to the vehicle model to generate a vehicle control amount with which the control target is achieved; and a prediction accuracy determination unit that calculates the accuracy of the prediction of the future behavior and changes, on the basis of the accuracy of the prediction, at least one of the model parameter or the vehicle model used in the prediction of the future behavior and the generation of the vehicle control amount.

Description

Vehicle behavior control device 【0001】 The present invention relates to a vehicle behavior control device. 【0002】 Conventionally, as driving assistance for maintaining stable running of a vehicle, a vehicle control device has been proposed that adjusts the driving force, braking force, and steering angle for each driving wheel based on the vehicle state predicted using a vehicle model, and adds a yaw moment to stabilize the running. In this vehicle control device, in order to calculate an appropriate yaw moment, it is necessary to maintain high prediction accuracy of the vehicle model. 【0003】 On the other hand, for example, in Patent Document 1, in order to cope with changes in vehicle performance due to parts replacement or aging, a plurality of vehicle models with different model parameters are prepared, and the direction and magnitude of changes in vehicle performance are obtained from the difference in the calculation results, and a method of correcting the added yaw moment is disclosed. 【0004】 Japanese Patent Application Laid-Open No. 2014 - 148262 【0005】 The technique of Patent Document 1 addresses the reduction in accuracy caused by errors in model parameters within the vehicle model. However, since the vehicle model is a mathematical formulation approximating the actual vehicle motion, not only errors in model parameters but also a decrease in prediction accuracy due to approximation errors can occur. In that case, in the method described in Patent Document 1, there remains room for improvement in running stability in a situation where the approximation error of the vehicle model cannot be ignored, and since it is necessary to calculate using a plurality of vehicle models with different model parameters, it is difficult to use a vehicle model with few approximations and a large computational load. 【0006】A vehicle behavior control device according to an aspect of the present invention includes: a control target generation unit that generates a control target based on a driving operation amount; a vehicle state acquisition unit that acquires the vehicle state; an environment information acquisition unit that acquires vehicle environment information; a model parameter estimation unit that estimates model parameters relating to a vehicle model based on the vehicle state and the vehicle environment information; a future behavior prediction unit that applies the vehicle state and the model parameters to a vehicle model to predict the future behavior of the vehicle; a vehicle control amount generation unit that applies the generated control target and the predicted future behavior to a vehicle model to generate a vehicle control amount that achieves the control target; and a prediction accuracy determination unit that calculates the prediction accuracy of the future behavior and changes at least one of the model parameters and the vehicle model in the prediction of the future behavior and the generation of the vehicle control amount based on the prediction accuracy. 【0007】 According to the present invention, the vehicle model and model parameters can be changed according to factors that reduce the accuracy of future behavior predictions, thereby improving driving stability. 【0008】 Figure 1 is a plan view showing the overall configuration of a vehicle to which the vehicle behavior control device of this embodiment is applied. Figure 2 is a block diagram showing an example of the configuration of the vehicle control device. Figure 3 is a diagram illustrating the future behavior calculated by the future behavior prediction unit. Figure 4 is a diagram showing the changes in the vehicle control quantity corresponding to the future behavior shown in Figure 3. Figure 5 is a diagram showing a vehicle model and an example of a tire model used within that vehicle model. Figure 6 is a diagram illustrating the tire model. Figure 7 is a diagram illustrating the prediction of future behavior. Figure 8 is a diagram illustrating the time required for future behavior prediction. Figure 9 is a diagram illustrating a method for reducing the number n of prediction time steps Se. Figure 10 is a flowchart showing an example of the prediction accuracy determination process. Figure 11 is a flowchart showing the detailed processing of step S10 in Figure 10. Figure 12 is a flowchart showing the detailed processing of step S20 in Figure 10. It is a diagram illustrating the prediction accuracy determination process. 【0009】The embodiments for carrying out the present invention will be described below with reference to the figures. The following description and drawings are illustrative examples for explaining the present invention, and have been omitted and simplified as appropriate for clarity of explanation. In addition, in the following description, the same or similar elements and processes are denoted by the same reference numerals, and redundant explanations may be omitted. It should be noted that the contents described below are merely examples of embodiments of the present invention, and the present invention is not limited to the embodiments described below, and can be carried out in various other forms. 【0010】 Figure 1 is a plan view showing the overall configuration of a vehicle 1 to which the vehicle behavior control device of this embodiment is applied. The vehicle 1 includes a vehicle control device 2, an external control device 3, a combine sensor 4, wheels 11 (11FL, 11FR, 11RL, 11RR), motors 12 (12FL, 12FR, 12RL, 12RR), brake mechanisms 13 (13aFL, 13aFR, 13aRL, 13aRR, 13b), steering mechanisms 14 (14F, 14R, 14aF, 14aR, 14bF, 14bR), suspension 15, accelerator pedal 16, brake pedal 17, steering wheel 18, external sensor 19, etc. 【0011】 In the symbols shown in Figure 1, the alphabetical symbols FL, FR, RL, and RR indicate the configuration corresponding to the left front, right front, left rear, and right rear, respectively. For example, for wheel 11, symbols 11FL, 11FR, 11RL, and 11RR represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. Additionally, the alphabetical symbols F and R indicate the configuration corresponding to the front and rear sides, respectively. Below, we define the longitudinal direction of vehicle 1 as the x-axis (forward direction is positive), the lateral direction as the y-axis (left direction is positive), and the vertical direction as the z-axis (upward direction is positive), and then we will sequentially explain the details of each configuration. 【0012】The vehicle control device 2 functions as a vehicle behavior control device in this embodiment, and integrates control of various actuators such as the motor 12, brake mechanism 13, steering mechanism 14, and suspension 15 in response to the driver's operation, external commands from the external control device 3, and detection signals from the combine sensor 4. The combine sensor 4 outputs detection signals related to a total of six control axes: longitudinal, lateral, and vertical acceleration, and roll, pitch, and yaw rates. 【0013】 The vehicle control device 2 is specifically an ECU (Electronic Control Unit) equipped with hardware such as a CPU or other arithmetic unit, a main memory such as semiconductor memory, an auxiliary memory, and a communication device. The arithmetic unit then executes a program loaded from the auxiliary memory into the main memory, thereby realizing the functions described later. In the following explanation, such well-known technologies will be omitted as appropriate. 【0014】 The external control unit 3 is a higher-level controller that performs driver assistance control and autonomous driving control via the vehicle control unit 2. Based on external information acquired by external sensors 19, which consist of cameras, radar, LiDAR, etc., the external control unit 3 calculates speed command values ​​and acceleration command values ​​for implementing adaptive cruise control (ACC) that follows the preceding vehicle, or yaw command values ​​for implementing lane keeping control (LKC) that maintains driving within the lane, and outputs these as external commands to the vehicle control unit 2. In the configuration shown in Figure 1, the vehicle control unit 2 receives the signals output from the external sensors 19, and these received signals are input to the external control unit 3. In Figure 1, the vehicle control unit 2 and the external control unit 3 are shown as separate units, but they may also be implemented in a single ECU. 【0015】As external sensors 19, for example, by installing fisheye cameras with a 180° field of view on the front (19F), left and right sides (19SL, 19SR), and rear (19R) of the vehicle 1, the relative distance and relative speed of objects such as other vehicles, bicycles, pedestrians, and obstacles in the vicinity of the vehicle 1 can be detected. Note that the external sensors 19 are not limited to the fisheye cameras in the above example, but may also be ultrasonic sensors, stereo cameras, infrared cameras, laser radars, or combinations thereof, or a laser radar capable of sensing a 360° surrounding area may be mounted on the roof of the vehicle 1. Signals output from these sensors are input to the vehicle control device 2 or the external control device 3. 【0016】 (Explanation of the drive system) Vehicle 1 is equipped with a torque generating device as the main part of its drive system, which provides driving force to each wheel 11. An example of a torque generating device is an engine or motor that transmits driving force to a pair of left and right wheels 11 via a differential gear and drive shaft. Another example of a torque generating device is an in-wheel motor that independently drives each wheel 11. In the example shown in Figure 1, each wheel 11 is equipped with an in-wheel motor 12 (12FL, 12FR, 12RL, 12RR). 【0017】 When the driver wants to move vehicle 1 forward (or backward), the driver sets the shift lever to the desired setting and then operates the accelerator pedal 16. The amount of depression of the accelerator pedal 16 is detected by the stroke sensor 16a. The acceleration control device 16b converts the detected depression amount into an accelerator command and outputs it to the vehicle control device 2. The vehicle control device 2 supplies power corresponding to the input accelerator command from a battery (not shown) to the motors 12 of each wheel and controls the torque of each motor. As a result, the vehicle 1 can be accelerated or decelerated in response to the operation of the accelerator pedal 16. 【0018】 Furthermore, when driving assistance or automated driving is performed in response to an external command from the external control device 3, the vehicle control device 2 supplies the desired power to the motors 12 of each wheel in accordance with the input external command, and controls the torque of each motor to accelerate or decelerate the vehicle 1. As a result, the desired driving assistance or automated driving is performed. 【0019】 (Explanation of the braking system) Each wheel 11 is equipped with a wheel cylinder 13a (13aFL, 13aFR, 13aRL, 13aRR) that applies braking force. The wheel cylinder 13a is composed of, for example, a cylinder, piston, pad, disc rotor, etc. In the wheel cylinder 13a, the piston is propelled by the working fluid supplied from the master cylinder, and the pad connected to the piston is pressed against the disc rotor, which rotates together with the wheel 11. As a result, the brake torque acting on the disc rotor becomes the braking force acting between the wheel 11 and the road surface. 【0020】 When the driver wants to brake vehicle 1, the driver operates the brake pedal 17. The force applied by the driver to the brake pedal 17 is increased by a brake booster (not shown), and the master cylinder generates hydraulic pressure proportional to the force applied. The generated hydraulic pressure is supplied to the wheel cylinders 13a (13aFL, 13aFR, 13aRL, 13aRR) of each wheel via the brake mechanism 13. As a result, the pistons of each wheel cylinder 13a are pressed against the disc rotor in response to the driver's brake pedal operation, and braking force is generated on each wheel 11. 【0021】 In addition, in a vehicle 1 equipped with the vehicle control device 2, the brake booster and master cylinder may be omitted. In that case, for example, the brake pedal 17 and the brake mechanism 13 are directly connected, so that when the driver presses the brake pedal 17, the brake mechanism 13 operates directly. 【0022】 Furthermore, when driving assistance or automated driving is performed in response to an external command from the external control device 3, the vehicle control device 2 controls the brake mechanism 13 and the wheel cylinders 13a of each wheel via the brake control device 13b in accordance with the input external command. As a result, the vehicle 1 is braked, and the desired driving assistance or automated driving is performed. The brake control device 13b also has the function of converting the amount of operation of the brake pedal 17 by the driver into a brake command and outputting it to the vehicle control device 2 as an external command. 【0023】(Explanation of the steering system) Vehicle 1 is equipped with a steering mechanism 14 that provides steering force to each wheel 11 as the main part of the steering system. In Figure 1, reference numeral 14F indicates the front steering mechanism that steers the front wheels 11F (left front wheel 11FL, right front wheel 11FR), and reference numeral 14R indicates the rear steering mechanism that steers the rear wheels 11R (left rear wheel 11RL, right rear wheel 11RR). Note that it is not necessary to have steering mechanisms 14 at both the front and rear; for example, the rear steering mechanism 14R may be omitted. 【0024】 When the driver wants to steer vehicle 1, the driver operates the steering wheel 18. The steering torque and steering angle input by the driver via the steering wheel 18 are detected by the steering torque detection device 18a and the steering angle detection device 18b. The front steering control device 14aF controls the front steering motor 14bF based on the detected steering torque and steering angle to generate assist torque for steering the front wheels 11F (11FL, 11FR). Similarly, the rear steering control device 14aR controls the rear steering motor 14bR based on the detected steering torque and steering angle to generate assist torque for steering the rear wheels 11R (11RL, 11RR). 【0025】 Furthermore, when driving assistance or automated driving is performed in response to an external command from the external control device 3, the vehicle control device 2 controls the steering torque of the steering motor 14b (14bF, 14bR) via the steering control device 14a (14aF, 14aR). As a result, the vehicle 1 is braked, and the desired driving assistance or automated driving is performed. In this case, the steering wheel 18 may be omitted. 【0026】(Explanation of the suspension system) Vehicle 1 is equipped with suspensions 15 (15FL, 15FR, 15RL, 15RR) as the main components of its suspension system. These suspensions absorb vibrations and shocks generated in each wheel 11, improving the stability of the vehicle and ride comfort. These suspensions 15 may include, for example, a semi-active suspension that combines a damper with adjustable viscosity and a coil spring, or a fully active suspension that combines an actuator with adjustable length, a damper, and a coil spring to arbitrarily change the relative distance between the vehicle body and the wheels 11. The vehicle control device 2 not only improves ride comfort by controlling the viscosity of the semi-active suspension and the length of the fully active suspension, but also appropriately controls the attitude of vehicle 1 according to the environment. 【0027】 <Description of Vehicle Control Device 2> Figure 2 is a block diagram showing an example of the configuration of the vehicle control device 2. The vehicle control device 2 includes at least a driver operation amount acquisition unit 20, a vehicle state acquisition unit 21, an external information acquisition unit 22, a model parameter estimation unit 23, a control target generation unit 24, a future behavior prediction unit 25, an information display unit 26, a vehicle control amount generation unit 27, and a prediction accuracy determination unit 28. 【0028】 (Driver Operation Amount Acquisition Unit 20) The driver operation amount acquisition unit 20 acquires the amount of driving operations performed by a driver who is seated in the vehicle 1 and performing driving operations. Examples of the driving operations acquired include the output value of the stroke sensor 16a when the accelerator pedal 16 is pressed, the output value of a switch or pedal force sensor that determines whether the brake pedal 17 is pressed, and the output values ​​of the steering torque detection device 18a and steering angle detection device 18b that acquire how much force and angle the steering wheel 18 is operated with. 【0029】(Vehicle State Acquisition Unit 21) The vehicle state acquisition unit 21 acquires vehicle states such as translational and rotational speed and acceleration occurring in the vehicle 1. For example, it may acquire detected values ​​of longitudinal, lateral, and vertical acceleration, as well as roll, pitch, and yaw rates from the combine sensor 4, or it may calculate the vehicle state from the operating amounts of each actuator such as the motor 12, brake mechanism 13, and steering mechanism 14. Furthermore, it may acquire detected values ​​of roll and pitch angles from information of the stroke sensors of the suspension 15 attached to each wheel, or it may be equipped with a GNSS (Global Navigation Satellite Systems) and acquire its own position (vehicle position) from the GNSS. The vehicle state acquired by the vehicle state acquisition unit 21 is stored in a storage device (RAM, etc.) in the vehicle control device 2. 【0030】 (External Information Acquisition Unit 22) The external information acquisition unit 22 acquires external information from external sensors 19 installed on the vehicle 1, and determines the position, size, and speed of objects (obstacles, etc.) around the vehicle 1 based on the acquired external information. It may also detect road signs, road markings, traffic lights, etc., and determine their position and type to recognize traffic rules in the area in front of the vehicle. When a camera is used as the external sensor 19, it is possible to acquire external information by simultaneously identifying the type of multiple objects from the image data of the camera. In particular, a stereo camera using two cameras can also detect the relative distance and relative speed of moving objects and obstacles. The type of road surface, moisture content, and freezing state may also be estimated from the color information of the road surface in the image. The external information acquired by the external information acquisition unit 22 is stored in a memory device (RAM, etc.) in the vehicle control device 2. 【0031】Furthermore, the system may include a map information acquisition unit (not shown) that acquires map information, such as the shape of the road on which the vehicle 1 travels and the location of road markings, in advance through storage or communication. In this case, the vehicle status acquisition unit 21 can acquire its own position with higher accuracy by correcting its own position by comparing the map information with information such as road markings in the external information acquisition unit 22. In addition, the external information acquisition unit 22 can acquire more accurate traffic rules from the map information. 【0032】 (Model Parameter Estimation Unit 23) The model parameter estimation unit 23 estimates the model parameters of a vehicle model that represents vehicle motion using the vehicle state and external information. The vehicle model itself may be corrected using a Kalman filter, or the vehicle mass and road surface friction coefficient used in calculating vehicle motion may be directly estimated. Alternatively, the estimation method may be performed using a pre-trained machine learning model. For example, the machine learning model may calculate parameter values ​​using a neural network model with the vehicle state and external information as input, or it may select from pre-prepared parameter values ​​using a random forest or support vector machine. The model parameters estimated by the model parameter estimation unit 23 are stored in a memory device (RAM, etc.) within the vehicle control device 2. 【0033】 (Control Target Generation Unit 24) The control target generation unit 24 calculates the target vehicle state from the driving operations acquired by the driver operation amount acquisition unit 20. For example, using a pre-prepared correspondence table, it sets the target acceleration and target speed from the operation amounts of the accelerator pedal and brake pedal, or sets the target yaw rate, target turning radius, and target trajectory from the steering torque and steering angle. For example, a vehicle model with characteristics desired by the driver may be stored in advance, and the vehicle state output when the driver operation amount is input may be set as the target value using that vehicle model. 【0034】(Future Behavior Prediction Unit 25) The future behavior prediction unit 25 uses the vehicle model determined by the prediction accuracy determination unit 28 and the model parameters estimated by the model parameter estimation unit 23 to calculate the future vehicle state from the current time t0 to time t1, which is a predicted time length ΔT ahead, as shown in Figure 3, as the future behavior (dashed line L10). In this case, the vehicle state acquired by the vehicle state acquisition unit 21 is set to the value (initial value) at the current time t0 in Figure 3. The vehicle control amount input to the vehicle model when calculating the future vehicle state up to time t1 is either a fixed value set in advance, or the vehicle control amount input to the vehicle 1 from the vehicle control amount generation unit 27, which will be described later, in the previous control cycle. Line L0 shows an example of a control target. 【0035】 For example, if the calculation period (control period) of the vehicle control amount output from the vehicle control amount generation unit 27 is set to 0.1 seconds, the predicted time length ΔT is set to approximately 2 seconds. That is, the future behavior prediction unit 25 calculates the vehicle state within the range of the predicted time length ΔT every 0.1 seconds. Figure 4 shows the transition of the vehicle control amount corresponding to the future behavior shown by the dashed line L10 in Figure 3, indicated by the dashed line L11. 【0036】 Figure 5 shows an example of a vehicle model used in the future behavior prediction unit 25 and a tire model used within that vehicle model. The vehicle model includes a 6-degree-of-freedom model that considers the vehicle's posture, including the center of gravity height, roll angle, and pitch angle, and a simplified 3-degree-of-freedom planar model that approximates it. The 6-degree-of-freedom model allows for the calculation of the load on each tire, enabling predictions that take into account changes in the maximum force that each tire can exert due to load changes. 【0037】 On the other hand, the planar three-degree-of-freedom model is a model that approximates roll angle and pitch angle to zero, assuming that fluctuations in these angles are negligibly small. While it does not consider changes in the load on each tire, the calculation load is reduced because roll angle and pitch angle are excluded from the calculated vehicle state. The maximum values ​​of roll angle φ and pitch angle θ to which the planar three-degree-of-freedom model can be applied are set as φmax and θmax, respectively. For example, φmax and θmax may be set based on simulations or experiments as limit values ​​where the prediction error of the vehicle state by the planar three-degree-of-freedom model makes it impossible to stabilize the vehicle's movement. 【0038】 The tire model, as shown in Figure 6, represents the magnitude of the tire lateral force Fy, which changes according to the tire slip angle β. The limit of the tire lateral force Fy is determined by the road surface friction coefficient, load, and tire wear, and the characteristics become more nonlinear as the tire slip angle β increases. In Figure 6, the solid line L21 shows the case of a nonlinear tire model that represents the nonlinear characteristics, and the dashed line L22 shows the case of a linear tire model that is a linear approximation. The linear tire model can be applied in the range below the slip angle βmax, where nonlinear characteristics appear. 【0039】 In the future behavior prediction unit 25, as shown in Figure 7, the predicted time length ΔT, which is the time from the current time t0 to the future time t1, is divided into multiple predicted time steps Se. Then, the vehicle state for each divided predicted time step Se, that is, the vehicle state at each timing shown by the black circles in Figure 7, is calculated. The set of multiple vehicle states calculated for each predicted time step Se in the predicted time length ΔT represents the future behavior in the predicted time length ΔT. 【0040】 Figure 8 illustrates the time required to calculate the future behavior prediction in each control cycle (hereinafter referred to as the required time Δt), and the required time Δt is shown by bar graphs G1 to G3. Bar graphs G1 and G3 show the required time Δt when using an unapproximated vehicle model, for example, as shown in Figure 4, number 4 (6-degree-of-freedom model, nonlinear). On the other hand, bar graph G2 shows the required time Δt when using an approximated vehicle model, for example, as shown in Figure 4, number 1 (planar 3-degree-of-freedom model, linear). Δj (j=1,2,3) is the calculation time required to calculate the vehicle state at each prediction time step Se. The number of prediction time steps Se n is n=5 for bar graphs G2 and G3, and n=3 for bar graph G1. 【0041】In non-approximated vehicle models, the calculation formula for calculating the vehicle state is more complex and includes nonlinear elements compared to approximated vehicle models, resulting in longer calculation times Δj at each predicted time step Se. For example, in bar graphs G2 and G3, the predicted time length ΔT for future behavior and the number of predicted time steps Se n are set to be the same. However, the calculation time Δj at each predicted time step Se differs depending on whether or not the model is approximated, with Δ2 < Δ3. 【0042】 By the way, since it is necessary to complete the calculation of future behavior within the control cycle, there is an upper limit Δtu for the required time Δt, which is the sum of the calculation times Δj. In the bar graph G3 shown in Figure 8, the required time Δt exceeds the upper limit Δtu, so it is necessary to make the required time Δt (= Δj・n) less than or equal to the upper limit Δtu. To do this, it is necessary to reduce the number n of predicted time steps Se in a non-approximated vehicle model. As a way to reduce the number n of predicted time steps Se, for example, the predicted time length ΔT may be shortened, or the time width of the predicted time step Se may be lengthened. 【0043】 Figure 9 shows an example of predicting future behavior when the predicted time length ΔT is shortened and when the predicted time step Se width is lengthened, given the settings for the predicted time length ΔT and the predicted time step Se width shown in Figure 7. When the predicted time is shortened, the predicted time step Se width is the same as in Figure 7, and the predicted time length ΔT1 is set to be three times the predicted time step Se width. When the predicted time step Se width is lengthened (increased), the predicted time length ΔT is the same as in Figure 7, and the predicted time step Se1 width is set to be longer than in Figure 7, to be one-third of the predicted time length ΔT. By setting it in this way, the required time Δt (=3Δ1) can be kept below the upper limit Δtu, as shown in bar graph G1 in Figure 8. 【0044】However, when the prediction time length ΔT is shortened, it becomes difficult to provide preventive driving support for early intervention. Also, when the time width of the prediction time step Se is increased, it becomes impossible to accurately predict future behavior, leading to a decline in control performance. In this embodiment, by using an approximated vehicle model and increasing the prediction time length ΔT while shortening the time width of the prediction time step Se, it becomes possible to provide preventive driving support for early intervention and to maintain control performance. 【0045】 (Information display unit 26) The information display unit 26 presents vehicle information to the driver by display. When unstable driving is predicted based on the future behavior (yaw angle at the vehicle center of gravity position) calculated by the future behavior prediction unit 25, the information display unit 26 causes display information indicating that unstable driving has been predicted to be displayed on a display device provided in the vehicle, for example. Since the driving becomes more unstable as the yaw angle at the vehicle center of gravity position indicating driving stability becomes larger, by notifying the driver of this fact by display when the preset threshold value is exceeded, it is possible to prompt the driver to perform an operation to stabilize the driving. Note that the notification of the occurrence of yaw is not limited to display, and a notification by voice or the like may be adopted, or they may be used in combination. 【0046】 (Vehicle control amount generation unit 27) The vehicle control amount generation unit 27 generates a vehicle control amount based on the future behavior calculated by the future behavior prediction unit 25 and the control target generated by the control target generation unit 24. For example, with respect to the difference between the future behavior L10 as shown in FIG. 3 and the control target, a vehicle control amount optimized based on the vehicle model may be generated so that the difference decreases, or a vehicle control amount optimized based on the vehicle model may be generated so as to suppress the yaw angle at the vehicle center of gravity position calculated from the future behavior to near zero. By generating the vehicle control amount so as to suppress the yaw angle at the future vehicle center of gravity position, it becomes possible to continue driving without impairing driving stability in the actual vehicle behavior. The vehicle control amount generated by the vehicle control amount generation unit 27 is stored in a storage device (such as a RAM) in the vehicle control device 2. 【0047】Note that, as the vehicle control amount, the driver operation amount corrected by calculating the steering angle, the accelerator pedal operation amount, and the brake pedal operation amount may be used, or the control amount for calculating the acceleration and deceleration speeds of each tire and driving the motor 12 and the brake mechanism 13 via the acceleration control device 16b and the braking control device 13b may be used. Further, as the vehicle model, the vehicle model determined as described later by the prediction accuracy determination unit 28 is used. 【0048】 (Prediction accuracy determination unit 28) The prediction accuracy determination unit 28 determines the prediction accuracy when using the vehicle model and the model parameters based on the vehicle state acquired by the vehicle state acquisition unit 21. When it is determined that the prediction accuracy has decreased, the vehicle model and the model parameters are changed. FIGS. 10 to 12 are flowcharts showing an example of the processing of the prediction accuracy determination unit 28. FIG. 10 shows the main processing, FIG. 11 shows the detailed processing of step S10 in FIG. 10, and FIG. 12 shows the detailed processing of step S20 in FIG. 10. 【0049】 In step S10 of FIG. 10, the prediction accuracy of the vehicle model is determined, and if the prediction accuracy is low, the vehicle model is changed. Next, in step S20, the prediction accuracy of the model parameters is determined, and if the prediction accuracy is low, the model parameters are changed. The factors causing the accuracy degradation include the approximation error of the vehicle model and the estimation error of the model parameters estimated by the model parameter estimation unit 23. By the processing of steps S10 and S20, reduction of the approximation error by switching the vehicle model and suppression of the degradation of the control performance by correcting the model parameters are achieved. Thus, when predicting the future behavior at the current time, it is determined whether the vehicle model and the model parameters are factors causing the accuracy degradation, and appropriate driving support becomes possible by implementing countermeasures for each of them. 【0050】Referring to Figure 11, the details of the process in step S10 of Figure 10 will be explained. In step S101 of Figure 11, the prediction accuracy determination unit 28 acquires the roll angle, pitch angle, and sideslip angle of each tire as vehicle conditions in order to select a vehicle model from the vehicle models shown in Figure 5 that can be applied to predicting future behavior from the current time. Here, in order to determine whether a vehicle model is applicable, starting with the vehicle model number 1 shown in Figure 5, the determination is made in order. Therefore, in step S102, an initial value of 1 is assigned to the variable k, and then in step S103, the applicability range of the k-th vehicle model from Figure 5 is obtained. 【0051】 In step S104, the acquired vehicle state is compared with the applicable range to determine whether the vehicle state satisfies the conditions of the applicable range shown in Figure 5. If it is determined in step S104 that it is not applicable (No), the process proceeds to step S105, where 1 is added to the variable k, and then the process returns to step S103. On the other hand, if it is determined in step S104 that it is applicable (Yes), the process proceeds to step S106, where the k-th vehicle model determined to be applicable is applied to the vehicle model used by the future behavior prediction unit 25 and the vehicle control amount generation unit 27. 【0052】 Furthermore, in comparing the vehicle condition and the applicable range as described above, in step S101, the roll angular velocity, pitch angular velocity, and lateral slip angular velocity of each tire may be obtained, and the roll angle, pitch angle, and lateral slip angle of each tire after a predicted time length may be calculated based on their respective rate of change and compared with the applicable range. 【0053】 As described above, by determining whether the vehicle models can be applied to predict future behavior from the current time, starting with those with low computational load, as in the processing steps S103 to S105, it is possible to prevent a decrease in prediction accuracy due to approximation errors in the vehicle models and to enhance the effectiveness of driver assistance through prediction. Furthermore, in situations where prediction accuracy does not decrease, it is possible to reduce the computational load and lower the power consumption required for calculation by using an approximated, simplified vehicle model. 【0054】Referring to Figure 12, the details of the process in step S20 of Figure 10 will be explained. In step S201 of Figure 12, the prediction accuracy determination unit 28 acquires past vehicle states and vehicle control quantities from the current time stored in the memory device in order to determine the prediction accuracy based on the vehicle model and model parameters based on the estimation error for past vehicle behavior. Here, as shown in Figure 13, data from the estimated initial time, which is set back by the estimated time length from the current time t0, to the current time t0 is acquired. In step S202, the vehicle state acquired at the estimated initial time is set as the initial value, and the vehicle control quantities for the estimated time length stored in the memory device are input to the vehicle model applied in step S106 of Figure 11, and the vehicle state from the estimated initial time to the current time t0 is calculated. Hereafter, the calculated vehicle state will be referred to as the estimated vehicle state. 【0055】 In step S203, the vehicle state at the estimated time length obtained in step S201 is compared with the estimated vehicle state calculated in step S202, and the estimation error is calculated. The estimation error may be either the mean squared error or the maximum error. 【0056】 In step S204, it is determined whether the estimation error calculated in step S203 is less than a predetermined threshold. Since vehicle motion changes continuously with respect to time, it can be assumed that the accuracy of the vehicle model's estimation of the vehicle state around the current time will remain the same even beyond the current time. The threshold here is the limit at which the vehicle's movement can no longer be stabilized due to the prediction error, and is set, for example, based on simulations or experiments. 【0057】 If it is determined in step S204 that the estimation error is below the threshold (Yes), the prediction accuracy is considered sufficient for driving stabilization, and the process is terminated without changing the model parameters. Here, the vehicle model used in the process of step S20 in Figure 10 has been changed to an applicable vehicle model in step S10, so the approximation error of the vehicle model is not considered to be a factor in reducing accuracy. Therefore, the estimation error calculated in step S203 can be considered to be an error caused by the model parameters. 【0058】On the other hand, if the estimation error is determined to be above a threshold (No) in step S204, the estimated vehicle state calculated in step S202 is considered to have a prediction accuracy that does not allow for stabilizing the vehicle's movement. If No is determined in step S204, that is, if the estimated vehicle state has a prediction accuracy that does not allow for stabilizing the vehicle's movement, the process in step S205 is executed so that driving assistance is provided assuming a situation where stabilizing the vehicle's movement is difficult. 【0059】 In step S205, the model parameters are changed to model parameters that make it difficult to stabilize the vehicle's driving, that is, model parameters that reduce the responsiveness of the vehicle state to the vehicle control amount. For example, assuming that the inertia is large relative to the vehicle control amount, the vehicle mass is set to a large value. Also, assuming that the force generated by the tires is small relative to the steering angle and tire torque, the road surface friction coefficient is set to a small value or to the characteristics of worn tires. In this way, by providing driving assistance that assumes a situation where it is difficult to stabilize the vehicle's driving when the prediction accuracy is reduced, it becomes possible to stabilize the vehicle's driving even if the estimation error of the model parameters acts in a way that degrades the control performance. 【0060】 According to the embodiments of the present invention described above, the following effects and advantages are achieved. 【0061】 (1) As shown in Figures 2, 10 to 12, etc., the vehicle control device 2 includes a control target generation unit 24 that generates control targets based on driving operation amounts, a vehicle state acquisition unit 21 that acquires the vehicle state, an external information acquisition unit 22 that acts as an environment information acquisition unit to acquire vehicle environment information (external information), a model parameter estimation unit 23 that estimates model parameters related to the vehicle model based on the vehicle state and external information, a future behavior prediction unit 25 that applies the vehicle state and model parameters to the vehicle model to predict the future behavior of the vehicle, a vehicle control amount generation unit 27 that applies the generated control targets and predicted future behavior to the vehicle model to generate vehicle control amounts that achieve the control targets, and a prediction accuracy determination unit 28 that calculates the prediction accuracy of future behavior and changes at least one of the model parameters and the vehicle model in the prediction of future behavior and the generation of vehicle control amounts based on the prediction accuracy. 【0062】As described above, the accuracy of predicting future behavior is calculated, and at least one of the model parameters and the vehicle model is modified based on that prediction accuracy. This allows for appropriate countermeasures against both the approximation error of the vehicle model and the estimation error of the model parameters. As a result, appropriate driving assistance is possible to stabilize the vehicle's movement. 【0063】 (2) In (1) above, as shown in Figures 10 to 12, the prediction accuracy determination unit 28 determines whether the decrease in prediction accuracy is due to the model parameters or the vehicle model. If the decrease in accuracy is due to the model parameters, the model parameters are changed. If the decrease in accuracy is due to the vehicle model, the vehicle model is changed. By changing the vehicle model or model parameters that are causing the decrease in accuracy, the vehicle's operation can be stabilized by taking appropriate measures for both the approximation error of the vehicle model and the estimation error of the model parameters. 【0064】 (3) In (2) above, as shown in Figures 10, 11, etc., if the acquired vehicle state is outside the applicable range of the vehicle model, the prediction accuracy determination unit 28 changes the vehicle model to a vehicle model in which the vehicle state is within the applicable range. In this way, when the vehicle model is a factor in reducing accuracy, by changing to a vehicle model in which the vehicle state is within the applicable range, that is, a vehicle model with fewer approximations, the number of calculations in the prediction can be reduced, such as shortening the prediction time, and the decrease in prediction accuracy can be suppressed without increasing the calculation time. 【0065】 (4) In (2) above, as shown in Figures 10 to 13, the prediction accuracy determination unit 28 calculates the estimated vehicle state in the estimated time period based on the vehicle state (initial value) acquired at the start of a predetermined period (estimated time period) that goes back in time from the calculation time (current time t0) and the vehicle control amount generated in the estimated time period. If the error between the calculated estimated vehicle state and the vehicle state acquired in the estimated time period is greater than or equal to a predetermined value (threshold), the model parameters are changed to model parameters that have a lower response of the vehicle state to the vehicle control amount (step S205). 【0066】As described above, if model parameters become a factor in reducing accuracy, the model parameters are set to reflect a situation where the responsiveness of vehicle motion is reduced. This enables driver assistance that stabilizes driving even if the estimation error of the model parameters acts in a way that destabilizes driving. 【0067】 (5) In (1) above, as shown in Figures 6 and 7, the future behavior prediction unit 25 sets multiple prediction time steps Se within a period from the start of prediction (current time t0) to a prediction time length ΔT ahead, and predicts future behavior by calculating the vehicle state at each of the multiple prediction time steps Se. The prediction time length ΔT and the prediction time steps Se are set so that the time required for prediction is less than or equal to a predetermined upper limit value Δtu. This makes it possible to use an approximate vehicle model and to lengthen the prediction time length ΔT while keeping the time width of the prediction time steps Se short, enabling preventive driving assistance that intervenes early, while maintaining control performance. 【0068】 (6) In (1) above, as shown in Figure 2, the vehicle control amount generation unit 27 calculates the future lateral slip amount from the future behavior and generates a vehicle control amount that keeps the future lateral slip amount below a predetermined value. For example, it generates a vehicle control amount optimized based on the vehicle model so that the lateral slip angle at the vehicle's center of gravity position, calculated from the future behavior, is kept near zero. As a result, in actual vehicle behavior, it becomes possible to continue driving without impairing driving stability. 【0069】 (7) In (1) above, as shown in Figure 2, the vehicle is further provided with an information display unit 26 as an information notification unit that notifies the driver of vehicle information, and the vehicle control amount generation unit 27 notifies a warning when unstable driving is predicted from future behavior. For example, if the future skid angle calculated from future behavior exceeds a predetermined value, a warning is displayed on the information display unit 26. This prompts the driver to take driver actions to stabilize the driving. 【0070】The embodiments described above are merely examples, and the present invention is not limited to these, as long as the features of the invention are not impaired, and various modifications are included. Other embodiments conceivable within the scope of the technical idea of ​​the present invention are also included within the scope of the present invention. 【0071】 1...Vehicle, 2...Vehicle control device, 3...External control device, 11...Wheel, 12...Motor, 19...External sensor, 20...Driver operation amount acquisition unit, 21...Vehicle state acquisition unit, 22...External information acquisition unit, 23...Model parameter estimation unit, 24...Control target generation unit, 25...Future behavior prediction unit, 26...Information display unit, 27...Vehicle control amount generation unit, 28...Prediction accuracy determination unit, Se...Prediction time step, Δtu...Upper limit, ΔT...Prediction time length

Claims

1. A vehicle behavior control device comprising: a control target generation unit that generates a control target based on driving operation quantities; a vehicle state acquisition unit that acquires vehicle state; an environmental information acquisition unit that acquires vehicle environmental information; a model parameter estimation unit that estimates model parameters related to a vehicle model based on the vehicle state and the vehicle environmental information; a future behavior prediction unit that applies the vehicle state and the model parameters to a vehicle model to predict the future behavior of the vehicle; a vehicle control quantity generation unit that applies the generated control target and the predicted future behavior to a vehicle model to generate a vehicle control quantity that achieves the control target; and a prediction accuracy determination unit that calculates the prediction accuracy of the future behavior and changes at least one of the model parameters and the vehicle model in the prediction of the future behavior and the generation of the vehicle control quantity based on the prediction accuracy.

2. A vehicle behavior control device according to claim 1, wherein the prediction accuracy determination unit determines whether the decrease in prediction accuracy is due to the model parameters or the vehicle model, changes the model parameters if the decrease in accuracy is due to the model parameters, and changes the vehicle model if the decrease in accuracy is due to the vehicle model.

3. A vehicle behavior control device according to claim 2, wherein the prediction accuracy determination unit changes the vehicle model to a vehicle model in which the vehicle state is included if the acquired vehicle state is outside the applicable range of the vehicle model.

4. A vehicle behavior control device according to claim 2, wherein the prediction accuracy determination unit calculates an estimated vehicle state during a predetermined period based on the vehicle state acquired at the start of a predetermined period retrospectively from the time of calculation and the vehicle control amount generated during the predetermined period, and if the error between the estimated vehicle state and the vehicle state acquired during the predetermined period is greater than or equal to a predetermined value, the model parameter is changed to a model parameter that has a lower response of the vehicle state to the vehicle control amount.

5. A vehicle behavior control device according to claim 1, wherein the future behavior prediction unit sets a plurality of prediction time steps within a period from the start of prediction to a prediction time length ahead, calculates the vehicle state at each of the plurality of prediction time steps to predict the future behavior, and the prediction time length and the prediction time steps are set so that the time required for the prediction is less than or equal to a predetermined upper limit.

6. A vehicle behavior control device according to claim 1, wherein the vehicle control amount generation unit calculates a future lateral slip amount from the future behavior and generates a vehicle control amount that makes the future lateral slip amount less than or equal to a predetermined value.

7. A vehicle behavior control device according to claim 1, further comprising an information notification unit for notifying the driver of vehicle information, wherein the information notification unit notifies a warning when unstable driving is predicted from the future behavior.

Images