Methods for establishing a driver's lateral and longitudinal coupled behavior model

Abstract

This invention relates to a driver's lateral and longitudinal coupled behavior model, belonging to the field of automotive simulation testing technology. The purpose of this invention is to consider the coupling relationship between the driver's lateral and longitudinal behaviors, i.e., a driver behavior model that simultaneously controls speed and direction. Furthermore, the driving anticipation and decision-making process incorporates road curvature, allowing the driver to adjust vehicle speed appropriately based on road curvature to ensure cornering safety. The model building process of this invention involves: establishing a driver anticipation behavior module, a driver decision-making behavior module, a driver prediction behavior module, and a driver optimization behavior module. In the model building process, this invention treats the vehicle as a two-dimensional entity rather than a simple point mass, using rectangles to represent the outline of the vehicle for safe driving, which is more realistic.

Description

Technical Field

[0001] This invention belongs to the field of automotive simulation testing technology. Background Technology

[0002] In recent years, automobiles have become an important means of transportation, but this has also led to serious traffic safety problems, with tens of thousands of people losing their lives in traffic accidents every year. The safety and stability of a qualified car require extensive testing, which necessitates driver involvement. However, real-vehicle testing is time-consuming and labor-intensive; therefore, virtual testing has become a hot research topic in automotive testing and evaluation. When conducting virtual tests on relevant vehicle performance, a virtual driver drives the car, thus reflecting some performance issues. In this case, the virtual driver needs to simulate the driving behavior of a real driver using a driver model.

[0003] Traditional driver models often consider either the driver's steering control behavior or the driver's longitudinal speed control behavior, i.e., the driver's following behavior. The former aims to maintain a constant longitudinal speed, while the latter aims to maintain a constant direction of travel. However, both of these driver models have certain limitations in application. Summary of the Invention

[0004] The purpose of this invention is to consider the coupling relationship between the driver's lateral and longitudinal behaviors, that is, to establish a driver behavior model that can simultaneously control speed and direction. In addition, the driving preview decision-making process incorporates road curvature, and the driver adjusts the vehicle speed appropriately according to the road curvature to ensure the safety of curves.

[0005] The model establishment process of this invention is as follows: S1. Establish a driver anticipation behavior module After initially discretizing the lateral and longitudinal accelerations, the aiming time is set for each group of lateral and longitudinal accelerations. The motion trajectory prediction within the range divides the aiming time into m equal parts, i.e. Assuming that the vehicle's lateral and longitudinal accelerations and sideslip angle are constant during the preview time, the following equations are derived using the following formulas to describe the vehicle's state information during the preview time: (1) The vehicle's location information is then updated using the following formula: (2) in, and Let x and y be the coordinates of the vehicle's centroid in the geodetic coordinate system. and The lateral and longitudinal velocities of the vehicle in the vehicle coordinate system. and For the vehicle's lateral and longitudinal accelerations, Let ω be the vehicle's heading angle and ω be the vehicle's yaw rate. The sideslip angle is the vehicle's center of gravity. ; Safety constraints are divided into two parts: road boundary safety constraints and safe speed constraints at curves. The coordinates of the four vertices of the safety rectangle are calculated using the following formula based on the vehicle's center of gravity coordinates and heading angle: (3) in, and Let A be the x and y coordinates of vertex A of the safety rectangle, and so on. These are the horizontal and vertical coordinates of the vehicle's center of gravity, respectively. The vehicle's heading angle. and The length and width of the vehicle's safety rectangle are given. By using the coordinates of the four vertices and the centroid of the safety rectangle, the question of whether the vehicle collides with the road boundary can be transformed into a problem of the positional relationship between the rectangle and a simple polygon. If no collision occurs, the trajectory is retained; otherwise, the trajectory is deleted. Safe speed constraints at curves: When a vehicle is traveling at a curve, the following relationship exists: (4) Where F is the lateral force of the vehicle at the curve, V is the vehicle speed, and R is the radius of curvature of the road. The coefficient of friction of the road. The angle of inclination of the road; When the road slope angle is zero, a style coefficient is introduced. The speed safety constraints for vehicles traveling on curves can then be obtained as follows: (5) During vehicle operation, by calculating the real-time road curvature, the maximum safe speed that the driver believes the vehicle can travel on a curve can be obtained, and the vehicle speed can be adjusted accordingly to allow the vehicle to pass through the curve safely. S2. Establish a driver decision-making behavior module The feasible region is determined based on three aspects: driving stability, driving efficiency, and handling smoothness, and the following objective functions are established for each aspect: (6) in, The objective function representing driving smoothness, For the first in motion trajectory prediction The distance from the vehicle's center of gravity to the road centerline at each point; The objective function representing driving efficiency, The longitudinal acceleration of the vehicle; and Represents the smoothness of driving operation. and This represents the change in lateral and longitudinal acceleration, that is, the change in lateral and longitudinal acceleration in the current decision-making cycle compared to the previous decision-making cycle. By introducing the appropriate weighting coefficients, the comprehensive objective function is obtained as follows: (7) The driver's decision-making behavior can be realized by calculating the objective function, which determines the relevant vehicle state information that conforms to the real driver, and outputs this information to the driver control behavior model. S3. Establish a driver prediction behavior module A three-degree-of-freedom vehicle model was used as the inner model for nonlinear model predictive control. It mainly describes the lateral, longitudinal, and yaw dynamic characteristics. The equations for the three-degree-of-freedom vehicle model are as follows: (8) in, For the quality of the car, For the moment of inertia of yaw rotation, Let yaw rate be the vehicle's angular velocity. and These are the lateral and longitudinal velocities in the vehicle coordinate system. For the front wheel steering angle and the rear wheel steering angle, and These are the longitudinal tire force and the lateral tire force of the front wheels, respectively. and These are the longitudinal tire force of the rear wheels and the lateral tire force of the rear wheels, respectively. After simplification and formula derivation, the following vehicle dynamics formula is obtained: (9) in, and These are the lateral and longitudinal velocities in the vehicle coordinate system. It is the resultant force acting longitudinally on the vehicle tires. For the longitudinal acceleration of the vehicle, and These are the lateral stiffness of the front and rear wheels of the car, respectively. These are the distances from the car's center of gravity to the front and rear axles, respectively. The steering angle of the vehicle's front wheels; Based on the fundamental principles of nonlinear model predictive control, the first step is to predict the relevant vehicle state variables for future times based on the current vehicle state, i.e., to predict the time domain. Then, based on the relevant objective function, constraints, and reference input, the corresponding control quantity is calculated, i.e., the control time domain. The longitudinal velocity, lateral velocity, and yaw rate of the vehicle are selected as the state variables of the system, i.e. Longitudinal acceleration and front wheel steering angle are selected as the control variables of the system, i.e. Therefore, the system can be expressed in the form of differential equations, i.e. Discretize the above model and predict the relevant state variables of the vehicle at future times, taking a very small sampling time. After discretizing the system using Euler, we get: (10) go through The state equation of the system obtained after the prediction step is as follows: (13) The state update is achieved through the above formula, that is, based on the vehicle-related state variables at the current moment, the relevant variables at the future moment are predicted. S4. Establish a driver behavior optimization module By simulating the driver's optimal behavior using an objective function, the optimal control quantity is solved, and the following objective function is established: (14) in and It is the objective function for tracking the lateral and longitudinal velocities of the vehicle. It is an objective function concerning driver handling smoothness. and These are the predicted values ​​of the vehicle's lateral and longitudinal velocities, respectively. and These are the reference inputs for the vehicle's lateral and longitudinal velocities, respectively. To control the amount of change in the input, , and These are the weighting coefficients of the three objective functions, used to apply corresponding terminal constraints to the control input. (15) Solving the above objective function yields the optimal control sequence in the control time domain that minimizes the objective function value. The first value of the control sequence is then used as the output, while the longitudinal acceleration signal is converted into information about the vehicle's accelerator and brake pedals. S5. Establish driver neuromuscular module This phenomenon can be simulated using the following transfer function: (16) in, This is the time constant of the driver's neuromuscular response.

[0006] The beneficial effects of this invention are: 1. In establishing the driver's pre-aiming decision-making behavior model, this invention considers aspects such as driving efficiency, smooth operation, and stable driving. At the same time, when the vehicle enters a curve, it can make reasonable decisions about the vehicle speed based on the road curvature and the current vehicle speed, thus simulating the driver's deceleration behavior on a curve.

[0007] 2. In establishing the driver control behavior model, this invention considers the driver's lateral and longitudinal coupling behavior and uses nonlinear model predictive control to realize the driver's steering, driving and braking behavior of the vehicle during driving, which more effectively simulates the process of a real driver driving a vehicle.

[0008] 3. In the process of model building, this invention treats the vehicle as a two-dimensional entity rather than a simple point mass, and uses rectangles to represent the outline of the vehicle's safe driving, which is more realistic. Attached Figure Description

[0009] Figure 1 This is a principle block diagram of the method for modeling the lateral and longitudinal coupled behavior of drivers; Figure 2 Flowchart of driver anticipation decision-making behavior; Figure 3 This is a schematic diagram of road boundary constraints. Figure 4 This is a diagram illustrating driving on a curve. Figure 5 This is a schematic diagram of a three-degree-of-freedom vehicle model; Figure 6 This is a schematic diagram of the vehicle's driving trajectory; Figure 7 This is a schematic diagram showing the simulated trajectory of the dual-movement lane segment and the safe rectangular outline of the vehicle. Figure 8 This is a graph showing the longitudinal speed variation of the vehicle. Figure 9 This is a graph showing the change in the vehicle's longitudinal acceleration. Figure 10 This is a graph showing the change in lateral acceleration of the vehicle. Figure 11 This is a graph showing the steering wheel angle. Detailed Implementation

[0010] When overtaking or changing lanes, it is necessary to adjust the vehicle speed appropriately to ensure driving safety. Furthermore, the longitudinal and lateral speeds of a real vehicle are not completely decoupled variables; there is a certain coupling relationship between them. Therefore, this invention proposes a driver behavior model based on nonlinear model predictive control, considering the coupling relationship between the driver's lateral and longitudinal behaviors, enabling simultaneous speed and direction control. In addition, the driving prediction decision-making process incorporates road curvature, allowing the driver to adjust the vehicle speed appropriately based on the road curvature to ensure safety on curves.

[0011] To overcome the shortcomings of traditional driver models, this invention provides a modeling method for the lateral and longitudinal coupling behavior of drivers. This modeling method is implemented through the following steps: Step 1: Establish a driver anticipation behavior module During driving, drivers always look forward, a behavior known as driver anticipation. Based on the limitations of the vehicle's maneuverability and its current state, the feasible domain for the vehicle within the future anticipation time is calculated. This feasible domain also includes information such as the vehicle's speed and acceleration, and unsafe or unreasonable feasible domains are eliminated.

[0012] Step 2: Establish a driver decision-making behavior module Step 1 yields the feasible domain of the vehicle within the pre-aiming time. Evaluation indices are established, and the trajectories within these feasible thresholds are optimized to determine the more ideal trajectory. The relevant information, such as the vehicle's lateral and longitudinal velocities, corresponding to this trajectory are output.

[0013] Step 3: Establish a driver prediction behavior module Using the information obtained from the pre-aiming decision as a reference input for the driver's predictive behavior model, and based on the relevant information of the vehicle's current state, the model predicts the vehicle's state and position information for the next few moments, reflecting the driver's understanding of the vehicle's performance.

[0014] Step 4: Establish a driver behavior optimization module After obtaining the anticipation and prediction information, in order to make their prediction information consistent with their anticipation information, the driver needs to find the optimal control behavior, namely the longitudinal acceleration of the vehicle and the steering wheel angle, so that the vehicle can drive according to the state determined by the driver's anticipation decision, and convert the longitudinal acceleration signal into driving and braking information in order to control the controlled vehicle.

[0015] Step 5: Establish the driver's neuromuscular module The driving or braking information and steering wheel angle information obtained through the above steps need to be transmitted to the controlled vehicle through the driver's neuromuscular system. This results in a certain reaction delay, which is usually a pure time lag. The driver's delayed behavior is represented by a transfer function.

[0016] The following combination Figures 1 to 11 The present invention will be described in detail below: The driver's horizontal and vertical coupled behavior model is mainly implemented through the following modules: driver anticipation behavior module, driver decision-making behavior module, driver prediction behavior module, driver optimization behavior module, and driver neuromuscular module. It includes the driver's anticipation, decision-making, and execution processes when driving a vehicle.

[0017] The driver anticipation behavior module primarily describes the process by which a driver's gaze is always forward while driving, observing the feasible region of the road ahead and providing prerequisites for the driver to determine how to control the vehicle. The driver decision-making behavior module primarily describes how the driver determines the safe and legal optimal vehicle trajectory within the feasible region using relevant evaluation indicators, outputting the vehicle's state information corresponding to the trajectory. The driver prediction behavior module primarily describes the driver's prediction behavior regarding the vehicle's state and position at subsequent moments based on the current vehicle state, reflecting the driver's perception of the vehicle's performance. The driver optimization behavior module primarily describes the process by which the driver seeks the optimal vehicle control behavior, enabling the vehicle to travel in the manner determined by the anticipation decision, while simultaneously converting the longitudinal acceleration signal output by the model into a driving and braking signal that can be applied to the controlled vehicle. The driver neuromuscular module primarily describes the delayed neuromuscular behaviors inherent in the driver.

[0018] The specific steps of this implementation method are as follows: The basic principle block diagram of the driver model is as follows: Figure 1 As shown, the driver's pre-aiming decision-making behavior module is mainly used to simulate the driver's pre-aiming behavior when driving a car and to make decisions on relevant pre-aiming information, including mobility limits, trajectory prediction, safety constraints, legality constraints, and comprehensive performance evaluation modules; the driver's control behavior module is used to simulate the driver's operation behavior after the pre-aiming decision, including prediction behavior, optimization behavior, neuromuscular delay module, and controlled vehicle; the method includes the following steps.

[0019] Step 1: Establish a driver anticipation behavior module like Figure 2 As shown in the flowchart of the driver's anticipation decision-making behavior, the car itself has corresponding limitations in maneuverability; the maximum acceleration and deceleration that the vehicle can achieve is within a certain range, that is... , ,in and These represent the vehicle's lateral and longitudinal accelerations, respectively. The lateral and longitudinal accelerations are discretized appropriately to form a two-dimensional plane. Each node on this plane corresponds to a set of lateral and longitudinal accelerations, and the distances between each discrete point of the lateral and longitudinal accelerations are respectively... , Each set of lateral and longitudinal accelerations can be used to predict the vehicle's trajectory. The predicted trajectory also includes relevant information about other vehicle states, such as lateral and longitudinal velocities, accelerations, and vehicle position. After obtaining the corresponding predicted trajectory information, the safety and legality of each trajectory are assessed, and unsafe or illegal trajectories are deleted. Finally, the optimal lateral and longitudinal accelerations are determined through a comprehensive evaluation index, i.e., the points in the diagram. Then judge at this time. , Does it meet the required accuracy? If not, it needs to be determined at the current optimal lateral and longitudinal acceleration points. The nearby region is then discretized again, and the above process is repeated until the required accuracy is achieved.

[0020] After initially discretizing the lateral and longitudinal accelerations, the aiming time is set for each group of lateral and longitudinal accelerations. The motion trajectory prediction within the range divides the aiming time into m equal parts, i.e. Assuming that the vehicle's lateral and longitudinal accelerations and sideslip angle are constant during the preview time, the following equations are derived using the following formulas to describe the vehicle's state information during the preview time: (1) The vehicle's location information is then updated using the following formula: (2) in, and Let x and y be the coordinates of the vehicle's centroid in the geodetic coordinate system. and The lateral and longitudinal velocities of the vehicle in the vehicle coordinate system. and For the vehicle's lateral and longitudinal accelerations, Let ω be the vehicle's heading angle and ω be the vehicle's yaw rate. The sideslip angle is the vehicle's center of gravity. .

[0021] Based on the vehicle's current position and state information, the above formula can be used to predict the trajectory within a pre-aiming time for each set of discretized lateral and longitudinal accelerations. These trajectories also contain relevant vehicle state information. The current feasible driving region of the vehicle can be obtained through the above trajectory prediction. Next, unsafe and illegal trajectories within the feasible region need to be deleted.

[0022] First, safety constraints are imposed, mainly divided into two parts: road boundary safety constraints and safe speed constraints at curves. When a vehicle is traveling on the road, the driver maintains a certain distance from road boundaries or other obstacles to avoid collisions. This safe distance is represented by a rectangle and is called the vehicle's safety rectangle. Figure 3 As shown, based on the vehicle's predicted feasible region according to the above motion trajectory, a portion of it could cause the vehicle to drive off the road boundary, leading to a collision. Therefore, when implementing road boundary safety constraints, these feasible regions should be removed to ensure driving safety. The coordinates of the four vertices of the safety rectangle are calculated using the following formula based on the vehicle's centroid coordinates and heading angle: (3) in, and Let A be the x and y coordinates of vertex A of the safety rectangle, and so on. These are the horizontal and vertical coordinates of the vehicle's center of gravity, respectively. The vehicle's heading angle. and Let the length and width be the safety rectangle of the vehicle. By using the coordinates of the four vertices and the centroid of the safety rectangle, the question of whether the vehicle will collide with the road boundary can be transformed into a problem of the positional relationship between the rectangle and a simple polygon. If no collision occurs, the trajectory is retained; otherwise, the trajectory is deleted.

[0023] When a vehicle approaches a curve, speed is crucial. If the speed is too high, a real driver will slow down to ensure the vehicle safely navigates the curve. Therefore, safe speed limits at curves primarily reflect the driver's action of adjusting the vehicle speed to a safe range before entering a curve. Figure 4 As shown, when a vehicle is traveling on a curve, the following relationship exists: (4) Where F is the lateral force of the vehicle at the curve, V is the vehicle speed, and R is the radius of curvature of the road. The coefficient of friction of the road. The angle of inclination is the road's slope. The turning radius of a vehicle is approximately equal to the radius of curvature of the road. To prevent vehicles from overturning on curves, basic conditions for safe driving through curves must be met. Furthermore, different drivers, while ensuring safety, will drive at different speeds.

[0024] When the road slope angle is zero, a style coefficient is introduced. The speed safety constraints for vehicles traveling on curves can then be obtained as follows: (5) Therefore, by calculating the real-time road curvature during vehicle operation, the maximum safe speed that the driver perceives for the vehicle when driving through a curve can be obtained. This allows for corresponding speed adjustments to ensure the vehicle safely navigates the curve. These two safety constraints make the driver model more consistent with real-world driver characteristics.

[0025] After safety constraints are applied, the legality of the vehicle must also be constrained, primarily by setting legal speed limits. Specifically, the maximum longitudinal speed of the vehicle should not exceed the legal speed limit. If the vehicle speed exceeds the legal speed limit within the pre-aiming time, the trajectory of that movement will be deleted from the feasible region to achieve legality constraints.

[0026] Step 2: Establish a driver decision-making behavior module The feasible region information obtained from the aforementioned driver pre-aiming model still contains numerous possible vehicle trajectories, requiring decision-making to output information consistent with real drivers. This invention primarily makes decisions regarding the feasible region based on three aspects: driving stability, driving efficiency, and operational smoothness. Driving stability mainly describes the driver's actions to ensure a smooth vehicle trajectory, preventing swaying and driving close to the road centerline. Driving efficiency mainly describes the driver's actions to complete driving tasks at higher speeds while ensuring safety and legality; higher speeds result in higher driving efficiency. Operational smoothness mainly describes the driver's actions to avoid significant abrupt changes in vehicle operation. Based on these three aspects, the following objective functions are established: (6) in, The objective function representing driving smoothness, For the first in motion trajectory prediction The distance from the vehicle's center of gravity to the road centerline at each point; The objective function representing driving efficiency, The longitudinal acceleration of the vehicle; and Represents the smoothness of driving operation. and This represents the change in lateral and longitudinal acceleration, that is, the change in lateral and longitudinal acceleration in the current decision-making cycle compared to the previous decision-making cycle.

[0027] In the objective functions described above, the denominators are all fixed standard values. Introducing the appropriate weighting coefficients yields the comprehensive objective function as follows: (7) As can be seen from the objective function above, the smaller the value, the more it conforms to the actual driver behavior. By calculating the objective function, the driver's decision-making behavior can be realized, and the relevant vehicle state information that conforms to the actual driver can be determined. This information is then output to the driver control behavior model.

[0028] Step 3: Establish a driver prediction behavior module The driver predictive behavior model primarily employs a nonlinear model predictive control method. Because this driver model needs to control the vehicle simultaneously in both the lateral and longitudinal directions, a three-degree-of-freedom vehicle model is used as the internal model for nonlinear model predictive control. This model mainly describes the lateral, longitudinal, and yaw dynamics, meeting the basic requirements of this driver model. Figure 5 The schematic diagram of a three-degree-of-freedom vehicle model is shown. Based on the diagram, the following equations can be derived: (8) in, For the quality of the car, For the moment of inertia of yaw rotation, Let yaw rate be the vehicle's angular velocity. and These are the lateral and longitudinal velocities in the vehicle coordinate system. For the front wheel steering angle and the rear wheel steering angle, and These are the longitudinal tire force and the lateral tire force of the front wheels, respectively. and These are the longitudinal tire force of the rear wheels and the lateral tire force of the rear wheels, respectively.

[0029] To ensure the simplicity and convenience of the model while meeting the required specifications, the following vehicle dynamics formulas were obtained after appropriate simplification and derivation: (9) in, and These are the lateral and longitudinal velocities in the vehicle coordinate system. It is the resultant force acting longitudinally on the vehicle tires. For the longitudinal acceleration of the vehicle, and These are the lateral stiffness of the front and rear wheels of the car, respectively. These are the distances from the car's center of gravity to the front and rear axles, respectively. This refers to the steering angle of the vehicle's front wheels.

[0030] Based on the fundamental principles of nonlinear model predictive control, the first step is to predict the relevant vehicle state variables for future times based on the current vehicle state, i.e., to predict the time domain. Then, based on the relevant objective function, constraints, and reference input, the corresponding control quantity is calculated, i.e., the control time domain. The longitudinal velocity, lateral velocity, and yaw rate of the vehicle are selected as the state variables of the system, i.e. Longitudinal acceleration and front wheel steering angle are selected as the control variables of the system, i.e. Therefore, the system can be expressed in the form of a differential equation, i.e. By appropriately discretizing the above model, the relevant state variables of the vehicle at future moments are predicted using a very small sampling time. After discretizing the system using Euler, we get: (10)

[0031] go through After the prediction step, the system's state equation can be obtained as follows: (13) By updating the state using the above formula, the relevant variables for future moments can be predicted based on the vehicle's current state variables, reflecting the driver's understanding of the vehicle's performance.

[0032] Step 4: Establish a driver behavior optimization module Based on the predicted values ​​of relevant state variables obtained above, drivers always aim to achieve the best operational effect with the smallest possible amount of input, meaning the actual vehicle motion information must match the predicted information. Therefore, by simulating the driver's optimization behavior using an objective function, the optimal control quantity is solved, and the following objective function is established: (14) in and It is the objective function for tracking the lateral and longitudinal velocities of the vehicle. It is an objective function concerning driver handling smoothness. and These are the predicted values ​​of the vehicle's lateral and longitudinal velocities, respectively. and These are the reference inputs for the vehicle's lateral and longitudinal velocities, respectively. To control the changes in input, namely the changes in longitudinal acceleration and steering wheel angle, , and These are the weighting coefficients for the three objective functions.

[0033] To ensure smooth vehicle operation, appropriate terminal constraints need to be applied to the control variables, namely: (15) To achieve the same control effect as the driver's pre-aiming decision, the objective function is solved to obtain the optimal control sequence in the control time domain that minimizes the objective function. The first value of the control sequence is then used as the output. Simultaneously, the longitudinal acceleration signal is converted into information about the accelerator and brake pedals for subsequent control of the vehicle.

[0034] Step 5: Establish the driver's neuromuscular module The neuromuscular model of a driver is mainly designed to demonstrate the physiological limitations that exist when a driver performs a intended action. This delay can usually be considered a pure time lag, so the following transfer function is used to simulate this phenomenon: (16) in, This is the time constant of the driver's neuromuscular response.

[0035] Joint simulation verification All five steps described above were completed in Simulink within MATLAB, with B-Class from Carsim selected as the controlled vehicle. The relevant parameters of this vehicle are shown in Table 1. Table 1

[0036] To create a closed loop between Carsim and the driver model, corresponding input / output ports need to be set. Input ports include steering wheel angle, throttle control, and brake control; output ports include vehicle lateral and longitudinal speeds, yaw rate, heading angle, engine speed, and gear information. The relevant parameters for the driver model are set as shown in Table 2. Table 2

[0037] With the aforementioned parameter settings, the driver model and the controlled vehicle form a closed loop. The model is then simulated and verified using a dual lane-change scenario. Figure 6 The image shows a comparison between the vehicle's trajectory and the road centerline. It can be seen from the image that this driver model can effectively simulate the driver's steering behavior. Figure 7The diagram shows the simulated trajectory of a double-switching lane segment and the outline of a car's safety rectangle. The scales of the horizontal and vertical axes differ significantly, but the size of the solid rectangle corresponds to the actual safety rectangle of the car, and the entire driving process did not involve a collision with the road boundary, thus meeting driving safety standards. Figure 8 As shown, this is the longitudinal curve of the vehicle in the vehicle coordinate system. Combined with road condition information, it can be seen that initially, during the straight-line driving phase, the vehicle is accelerating. Approaching the first curve, the vehicle decelerates and passes through the curve at a safe speed. Similarly, at the following curves, the vehicle's speed is adjusted according to the road curvature. Finally, it enters the straight-line driving phase again, accelerating to maximum speed before ceasing acceleration. Figure 9 , 10 Figures 1 and 11 show the curves of vehicle longitudinal acceleration, vehicle lateral acceleration, and steering wheel angle, respectively. The joint simulation described above verifies the realism and effectiveness of the driver model.

[0038] Explanation of symbols involved in this invention: Vehicle longitudinal acceleration lateral acceleration of the vehicle , Maximum and minimum longitudinal acceleration of the vehicle , Maximum and minimum lateral acceleration of the vehicle , Dispersion of longitudinal and lateral acceleration Pre-aiming time Pre-aiming time Discrete time after equal division Time sequence after pre-aiming time is divided equally , The longitudinal and lateral velocities of the vehicle at time j-1 in the geodetic coordinate system. , : The lateral velocity of the vehicle at times j-1 and j in the vehicle coordinate system , The longitudinal velocity of the vehicle at times j-1 and j in the vehicle coordinate system. , Vehicle heading angles at times j-1 and j Vehicle yaw rate at time j Vehicle center of gravity sideslip angle The x-coordinates of the vehicle's centroid at times j-1 and j in the geodetic coordinate system. : The ordinate of the vehicle's centroid at times j-1 and j in the geodetic coordinate system The x-coordinates of the four vertices of the vehicle safety rectangle The ordinates of the four vertices of the vehicle safety rectangle The x and y coordinates of the vehicle's center of mass The length and width of the vehicle safety rectangle vehicle heading angle F: Lateral force of the vehicle at the curve V: Vehicle speed R: Radius of curvature of the road Road friction coefficient The angle of inclination of the road Cornering speed style coefficient Maximum safe speed when driving on curves Driving stability objective function : The distance from the vehicle's center of mass at point j to the road centerline in the motion trajectory prediction Driving efficiency objective function , Objective function for driving operation smoothness , Changes in lateral and longitudinal acceleration : Standard value of the corresponding objective function : Weight coefficients corresponding to the objective function Comprehensive objective function Overall vehicle weight yaw moment of inertia : Vehicle yaw rate Front wheel steering angle and rear wheel steering angle , Front wheel longitudinal tire force and lateral tire force , Rear wheel longitudinal tire force and lateral tire force The resultant force acting longitudinally on a vehicle's tires Vehicle longitudinal acceleration , Lateral stiffness of the front and rear wheels of a car Distance from the car's center of gravity to the front and rear axles Predicting the time domain Control Time Domain State variables Control variables Steering wheel angle Sampling time System state equations , Objective function for tracking vehicle lateral and longitudinal velocities Objective function for driver handling smoothness , Predicted values ​​of vehicle longitudinal and lateral speeds , Reference input for vehicle lateral and longitudinal velocities : Control the amount of change in the input , , : , , Weighting coefficients : Controls the maximum value of the input quantity Maximum steering wheel angle : The time constant of the driver's neuromuscular response.

Claims

1. A method for establishing a driver's lateral and longitudinal coupled behavior model, characterized in that: S1. Establish a driver anticipation behavior module After initially discretizing the lateral and longitudinal accelerations, the aiming time is set for each group of lateral and longitudinal accelerations. The motion trajectory prediction within the range divides the aiming time into m equal parts, i.e. Assuming that the vehicle's lateral and longitudinal accelerations and sideslip angle are constant during the preview time, the following equations are derived using the following formulas to describe the vehicle's state information during the preview time: （1） The vehicle's location information is then updated using the following formula: （2） in, and Let x and y be the coordinates of the vehicle's centroid in the geodetic coordinate system. , In geodetic coordinate system The vehicle's longitudinal and lateral speeds at any given time. and The lateral and longitudinal velocities of the vehicle in the vehicle coordinate system. The longitudinal acceleration of the vehicle, For the vehicle's lateral acceleration, The vehicle's heading angle. Let yaw rate be the vehicle's angular velocity. The sideslip angle is the vehicle's center of gravity. for The vehicle's yaw rate at any given moment. The time sequence after the pre-aiming time is divided into equal parts. , , In the vehicle coordinate system and The lateral speed of the vehicle at any given moment. , In the vehicle coordinate system and The longitudinal speed of the vehicle at any given moment; Safety constraints are divided into two parts: road boundary safety constraints and safe speed constraints at curves. The coordinates of the four vertices of the safety rectangle are calculated using the following formula based on the vehicle's center of gravity coordinates and heading angle: （3） in, and Vertices of a safe rectangle The x and y coordinates, and so on. These are the horizontal and vertical coordinates of the vehicle's center of gravity, respectively. The vehicle's heading angle. and The length and width of the vehicle's safety rectangle are given. By using the coordinates of the four vertices and the centroid of the safety rectangle, the question of whether the vehicle collides with the road boundary can be transformed into a problem of the positional relationship between the rectangle and a simple polygon. If no collision occurs, the trajectory is retained; otherwise, the trajectory is deleted. Safe speed constraints at curves: When a vehicle is traveling at a curve, the following relationship exists: （4） in, The lateral force exerted on the vehicle at a curve. For the vehicle's speed, Let be the radius of curvature of the road. The coefficient of friction of the road. The angle of inclination of the road. For the overall vehicle weight; When the road inclination angle is zero, a style coefficient is introduced. The speed safety constraints for vehicles traveling on curves can then be obtained as follows: （5） In the formula, This is the maximum safe speed for driving on a curve; During vehicle operation, by calculating the real-time road curvature, the maximum safe speed that the driver believes the vehicle can travel on a curve can be obtained, and the vehicle speed can be adjusted accordingly to allow the vehicle to pass through the curve safely. S2. Establish a driver decision-making behavior module The feasible region is determined based on three aspects: driving stability, driving efficiency, and handling smoothness, and the following objective functions are established for each aspect: （6） in, The objective function representing driving smoothness, For the first in motion trajectory prediction The distance from the vehicle's center of gravity to the road centerline at each point; The objective function representing driving efficiency, The longitudinal acceleration of the vehicle; and The objective function representing the smoothness of driving operation is... and This represents the change in lateral and longitudinal acceleration, specifically the change in lateral and longitudinal acceleration between the current decision-making cycle and the previous decision-making cycle. , , , These are the standard values ​​of the corresponding objective functions; By introducing the corresponding weighting coefficients, the comprehensive objective function is obtained as follows: （7） In the formula, For the comprehensive objective function, , , , These are the weight coefficients corresponding to the objective function; The driver's decision-making behavior can be realized by calculating the objective function, which determines the relevant vehicle state information that conforms to the real driver, and outputs this information to the driver control behavior model. S3. Establish a driver prediction behavior module A three-degree-of-freedom vehicle model was used as the inner model for nonlinear model predictive control. It mainly describes the lateral, longitudinal, and yaw dynamic characteristics. The equations for the three-degree-of-freedom vehicle model are as follows: （8） in, For the overall vehicle quality, For the moment of inertia of yaw rotation, Let yaw rate be the vehicle's angular velocity. and These are the lateral and longitudinal velocities in the vehicle coordinate system. and For the front wheel steering angle and the rear wheel steering angle, and These are the longitudinal tire force and the lateral tire force of the front wheels, respectively. and These are the longitudinal tire force of the rear wheels and the lateral tire force of the rear wheels, respectively. After simplification and formula derivation, the following vehicle dynamics formula is obtained: （9） in, and These are the lateral and longitudinal velocities in the vehicle coordinate system. It is the resultant force acting longitudinally on the vehicle tires. For the longitudinal acceleration of the vehicle, and These are the lateral stiffness of the front and rear wheels of the car, respectively. and These are the distances from the car's center of gravity to the front and rear axles, respectively. The steering angle of the vehicle's front wheels; Based on the fundamental principles of nonlinear model predictive control, the first step is to predict the relevant vehicle state variables for future times based on the current vehicle state, i.e., to predict the time domain. Then, based on the relevant objective function, constraints, and reference input, the corresponding control quantity is calculated, i.e., the control time domain. The longitudinal velocity, lateral velocity, and yaw rate of the vehicle are selected as the state variables of the system, i.e. Longitudinal acceleration and front wheel steering angle are selected as the control variables of the system, i.e. Therefore, the system can be expressed in the form of differential equations, i.e. , The system state equations are given; the above model is discretized, and the relevant state variables of the vehicle at future times are predicted by taking a very small sampling time. After discretizing the system using Euler, we get: （10） go through The state equation of the system obtained after the prediction step is as follows: （13） The state update is achieved through the above formula, that is, based on the vehicle-related state variables at the current moment, the relevant variables at the future moment are predicted. S4. Establish a driver behavior optimization module By simulating the driver's optimal behavior using an objective function, the optimal control quantity is solved, and the following objective function is established: （14） in and It is the objective function for tracking the lateral and longitudinal velocities of the vehicle. It is an objective function concerning driver handling smoothness. and These are the predicted values ​​of the vehicle's lateral and longitudinal velocities, respectively. and These are the reference inputs for the vehicle's lateral and longitudinal velocities, respectively. To control the amount of change in the input, , These are the predicted values ​​for the vehicle's longitudinal and lateral velocities, respectively. , and These are the weighting coefficients for the three objective functions, used to apply corresponding terminal constraints to the control input: （15） In the formula, To control the maximum value of the input, This represents the maximum steering wheel angle. Solving the above objective function yields the optimal control sequence in the control time domain that minimizes the objective function value. The first value of the control sequence is then used as the output, while the longitudinal acceleration signal is converted into information about the vehicle's accelerator and brake pedals. S5. Establish driver neuromuscular module This phenomenon can be simulated using the following transfer function: （16） in, This is the time constant of the driver's neuromuscular response.

Images