An intelligent vehicle lane-changing path planning method based on risk field

Abstract

The application discloses a kind of intelligent vehicle lane-changing path planning methods based on risk field, belongs to intelligent vehicle path planning field.The planning method includes the following steps: establishing the driving risk field including road boundary risk field, lane marking line risk field and obstacle risk field, while introducing vehicle size information and heading angle information to expand the applicable range of risk field;Then, combined with the vehicle kinematic model, the mass center horizontal and vertical coordinates, intelligent heading angle are used as state variables, and the front wheel rotation angle is used as control variable to establish model predictive controller to obtain optimal control variable, and according to the state variable obtained by model predictive controller, the calculation process is repeated at each sampling time to obtain dynamic vehicle path.The method of the application can more accurately and comprehensively describe the change of driving risk by establishing the expanded driving risk field.The driving environment in the future period of time is predicted by the path planning system model, and a more optimized control effect is realized.

Description

Technical Field

[0001] This invention belongs to the field of intelligent vehicle path planning, specifically relating to an intelligent vehicle lane-changing path planning method based on a risk field. Background Technology

[0002] Decision planning in autonomous driving technology is a key technology for realizing vehicle intelligence. Its main task is to plan the vehicle's trajectory based on environmental signals processed by the environmental perception system, while satisfying various constraints of vehicle driving, with the goal of optimizing a certain performance index.

[0003] Currently, trajectory planning algorithms mainly include graph search-based, curve fitting-based, mathematical optimization-based, artificial potential field-based, and sampling-based algorithms. As an effective means of driving safety assessment, the driving risk field considers the comprehensive impact of vehicle-road factors on driving safety, providing a new basis for judgment in driving risk assessment and active vehicle control under complex environments.

[0004] Examples include the invention patent applications "Method and System for Trajectory Planning of Connected Vehicles with Obstacle Avoidance and Lane Changing" (CN114537380A) and "Method and System for Intelligent Vehicle Trajectory Planning Based on Spatiotemporal Sampling Evaluation" (CN116149336A).

[0005] Among them, the Chinese invention patent application publication (CN114537380A), published on February 17, 2022, entitled "Method and System for Lane Changing Trajectory Planning for Connected Vehicles with Obstacle Avoidance," obtains the initial state of the main vehicle and the state information of surrounding vehicles, determines the transition state of the main vehicle for lane changing based on the state information, uses the initial state and transition state of the main vehicle as inputs to perform the first trajectory planning, determines whether the first trajectory planning meets safety and comfort constraints, and if it does, determines the lane changing end state based on the first trajectory planning; otherwise, it returns to the first trajectory planning. Using the transition state and lane changing end state of the main vehicle as inputs, it performs the second trajectory planning, determines whether the second trajectory planning meets safety and comfort constraints, and if it does, uses the second trajectory planning as the target lane changing trajectory; otherwise, it returns to the second trajectory planning. However, this method has the following shortcomings:

[0006] 1) The trajectory planning method does not take into account the size information of obstacles and the vehicle itself, which makes it impossible to accurately describe the lane-changing and obstacle-avoidance process of the vehicle.

[0007] The Chinese invention patent application publication (CN116149336A), published on February 24, 2023, entitled "Intelligent Vehicle Trajectory Planning Method and System Based on Spatiotemporal Sampling Evaluation," first initializes the planner; updates the road search space representation; generates optional trajectories starting from the vehicle's position and ending at nodes in the grid; generates optional trajectories using two nodes in the grid as the starting and ending points; selects the optimal ending point among the grid nodes; and backtracks to the vehicle's position from the optimal ending point to select the optimal trajectory. However, this method has the following shortcomings:

[0008] 1) The trajectory planning method using cubic curves can only guarantee the continuity of velocity and displacement, but not the continuity of acceleration, which affects the comfort of path planning.

[0009] 2) This path planning method makes extensive use of offline initial estimation tables as the initial solution for path generation parameters and ensures a certain path coverage, resulting in excessive computation and poor real-time performance. Summary of the Invention

[0010] The technical problem to be solved by this invention is the aforementioned technical shortcomings. This invention proposes an intelligent vehicle lane-changing path planning method based on a risk field. This method first establishes a driving risk field including a road boundary risk field, a lane marking risk field, and an obstacle risk field. Simultaneously, it introduces vehicle size information and heading angle information to expand the applicability of the risk field. Then, combining the vehicle kinematic model, it establishes a model predictive controller using the centroid's lateral and longitudinal coordinates, intelligent heading angle as state variables, and front wheel steering angle as control variables to achieve lane-changing path planning.

[0011] The objective of this invention is achieved as follows: This invention provides a method for intelligent vehicle lane-changing path planning based on a risk field, comprising the following steps:

[0012] Step 1, Establish the road boundary risk field E d_tq

[0013] The horizontal axis is set as the axis parallel to the horizon, and the positive direction is the front of the intelligent vehicle. The vertical axis is set as the axis perpendicular to the horizon in the ground plane, and the positive direction is the direction of rotating 90° counterclockwise from the front of the intelligent vehicle. The intersection of the two coordinate axes is the origin of the coordinate system. The road boundary and lane marking lines are parallel to the horizontal axis.

[0014] The risk distribution at the road boundary is assumed to exhibit a gradient descent trend, and the road boundary risk field E d_tq Described in power function form, its expression is:

[0015]

[0016] In the formula, μ is the road boundary field strength coefficient, and d tqLet d be the distance vector between the intelligent vehicle and the road boundary. tq d tq = (0, |y q -y t |-0.5(L sinθ+wcosθ)), the vehicle side distance vector d tq The direction starts from the centroid of the intelligent vehicle and points to the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle. t Let y be the ordinate of the centroid of the intelligent vehicle. q Let L be the ordinate of the centroid of the road boundary, which is on the same x-coordinate as the intelligent vehicle; w be the width of the intelligent vehicle; and θ be the heading angle of the intelligent vehicle.

[0017] Step 2, establish the lane marking risk field E b_tm

[0018] The lane marking risk field E b_tm The expression is:

[0019]

[0020] In the formula, D m The risk coefficient for lane marking lines, d tm Let d be the distance vector between the intelligent vehicle and the lane marking line m. tm , Vehicle line distance vector d tm The direction starts from the centroid of the intelligent vehicle and points to the centroid of the lane marking line on the same horizontal coordinate as the intelligent vehicle. m The ordinate of the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle;

[0021] Step 3, establish the obstacle risk field E z_ti

[0022] The obstacle risk field E z_ti The expression is:

[0023]

[0024] In the formula, λ is the first given constant, τ1 is the second given constant, and τ2 is the third given constant, all of which are greater than 0; b is the sequence number of the obstacle, b = 1, 2, meaning there are two types of obstacles, denoted as obstacle i. b Where obstacle i1 is an obstacle with negligible size, and obstacle i2 is an obstacle with non-negligible size; v ib For obstacle i b speed; d ti_b Let i be the distance vector between the intelligent vehicle and the obstacle. bThe distance vector; For obstacle i b The velocity direction and the distance vector d from the obstacle ti_b The angle between the two points is denoted as the velocity-distance angle. M ib For equivalent mass, characterize the intelligent vehicle and obstacle i b The severity of the collision;

[0025] Step 4, given the final objective function P of the lane-changing path planning.

[0026] Based on the road boundary risk field E d_tq Lane marking risk field E b_tm and obstacle risk field E z_ti By combining the vehicle kinematics model to establish a model predictive controller to realize lane-changing path planning, the final objective function P is obtained.

[0027] Step 4.1: Establish the vehicle kinematic model in the inertial coordinate system, its expression is:

[0028]

[0029] In the formula, Let be the derivative of the x-coordinate of the rear axle center of the intelligent vehicle. Let be the derivative of the longitudinal coordinate of the rear axle center of the intelligent vehicle. v is the derivative of the heading angle of the intelligent vehicle. tb Let ζ be the rear axle center velocity of the intelligent vehicle, ζ be the front wheel steering angle of the intelligent vehicle denoted as front wheel steering angle ζ, and γ be the wheelbase of the intelligent vehicle.

[0030] Step 4.2, based on step 4.1, set the x-coordinate of the intelligent vehicle's centroid to x. t The centroid ordinate y of the intelligent vehicle t The intelligent vehicle uses its heading angle θ as a state variable and the front wheel steering angle ζ as a control variable to establish a model predictive controller, the expression of which is:

[0031]

[0032] In the formula, e is the current sampling time, e+1 is the next sampling time, and W is the sampling period; Step 4.3, determine the final objective function P.

[0033] E of the road boundary risk field d_tq As the objective function P1, the lane marking risk field E b_tm As the objective function P2, the obstacle risk field E z_ti As the objective function P3:

[0034]

[0035] In the formula, U1 is the prediction time domain at each sampling time, j is the step size at the current sampling time, j = 1, 2, ..., U1, E d_tq (j) represents the road boundary risk field E at the current sampling time and the j-th prediction step. d_tq E b_tm (j) represents the lane marking risk field E at the j-th prediction step time at the current sampling moment. b_tm E z_ti (j) represents the obstacle risk field E at the j-th prediction step time at the current sampling moment. z_ti ;

[0036] The centroid of the intelligent vehicle is y-coordinated. t The error between the expected value and the target value is used as objective function P4, and the error between the intelligent vehicle's heading angle θ and the expected value is used as objective function P5, with the following expressions:

[0037]

[0038] In the formula, y t_w (j) represents the longitudinal position error of the intelligent vehicle at the j-th prediction step, θ w (j) represents the intelligent vehicle's heading angle error at the j-th prediction step, y t θ(j) represents the longitudinal position of the intelligent vehicle at the j-th prediction step, θ(j) represents the heading angle of the intelligent vehicle at the j-th prediction step, and y e (j) represents the expected longitudinal position of the intelligent vehicle at the j-th prediction step, θ e (j) represents the expected heading angle of the intelligent vehicle at the j-th prediction step;

[0039] Taking the front wheel steering angle ζ as the objective function P6, its expression is:

[0040]

[0041] In the formula, ζ(j) is the front wheel angle at the j-th prediction step, and U2 is the control time domain at each sampling time, satisfying U2≤U1;

[0042] Introducing the first weight coefficient ε1, the second weight coefficient ε2, the third weight coefficient ε3, the fourth weight coefficient ε4, the fifth weight coefficient ε5, and the sixth weight coefficient ε6, the final objective function P is expressed as follows:

[0043] P=ε1P1+ε2P2+ε3P3+ε4P4+ε5P5+ε6P6

[0044] Step 4.4, combining the vehicle steering structure saturation threshold, describes the model predictive control optimization problem of path planning, and obtains the optimal front wheel steering angle ζ.best Its expression is:

[0045]

[0046] In the formula, ζ min The minimum value of the front wheel steering angle ζ, ζ max This represents the maximum value of the front wheel steering angle ζ;

[0047] Step 5: At each sampling time, the optimal front wheel steering angle ζ is obtained by solving for the objective function P and the vehicle steering structure saturation threshold. best Then return to step 4.2 to obtain the centroid x-coordinate of the intelligent vehicle at sampling time e+1 based on the model predictive controller. t The centroid ordinate y of the intelligent vehicle t The heading angle θ of the intelligent vehicle t is calculated; the above process is repeated at each sampling moment during the lane-changing process of the intelligent vehicle to obtain a real-time optimized lane-changing path.

[0048] Preferably, the risk factor D of the lane marking line m The formula for calculation is:

[0049]

[0050] In the formula, T m Let d1 be the lane marking coefficient, and d1 be the lateral distance between the intelligent vehicle's centroid and the lane marking line m. d1 = |y t -y m |,y m Let w be the ordinate of the centroid of the lane marking line m, which shares the same x-coordinate as the centroid of the intelligent vehicle t. * This refers to the lane width.

[0051] Preferably, the equivalent mass M ib The formula for calculation is:

[0052]

[0053] In the formula, m ib For obstacle i b Actual quality, T ib For obstacle i b The type coefficient, a is the velocity polynomial The number of terms, ξ a Given the fourth constant, η a Given the fifth constant, λ a The sixth given constant;

[0054] The obstacle distance vector d ti_b The calculation is as follows:

[0055] For obstacle i1, the obstacle distance vector dti_b The formula for calculation is:

[0056]

[0057] In the formula, t1 is the distance between obstacle i1 and the center of mass of the intelligent vehicle. x t Let x1 be the x-coordinate of the intelligent vehicle's center of mass, x1 be the x-coordinate of obstacle i1, and y1 be the y-coordinate of obstacle i1. Let d be the velocity direction of obstacle i1 and the distance vector d between the vehicle and the obstacle. ti_b The included angle;

[0058] For obstacle i2, the vehicle-obstacle distance vector d ti_b The formula for calculation is:

[0059]

[0060] In the formula, t2 is the distance between the obstacle i2 and the centroid of the intelligent vehicle i. x2 is the x-coordinate of the centroid of obstacle vehicle i2, y2 is the y-coordinate of the centroid of obstacle vehicle i2, β1 is the length of obstacle vehicle i2, β2 is the width of obstacle vehicle i2, α1 is an undetermined coefficient related to the length of obstacle i2, α2 is an undetermined coefficient related to the width of obstacle i2, v2 is the velocity of obstacle i2, and Ω is an undetermined coefficient related to the velocity of obstacle i2. t Let y be the x-coordinate of the centroid of the intelligent vehicle. t Let t be the ordinate of the centroid of the intelligent vehicle. Let d be the velocity direction of the center of mass of obstacle i2 and the distance vector d between the vehicle and the obstacle. ti_b The included angle.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. By determining the obstacle type coefficient, this invention summarizes the distance information between the obstacle vehicle and surrounding vehicles with different heading angles, taking into account size, and establishes an expanded driving risk field that can more accurately and comprehensively describe the changes in driving risk.

[0063] 2. This invention uses model predictive control, with the front wheel angle as the control variable, to prevent the vehicle from turning too much and ensure comfort. By accurately modeling the path planning system model, the driving environment in the future is predicted, ensuring real-time performance and achieving a more optimized control effect. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating the implementation of the planning method of this invention.

[0065] Figure 2 This is a flowchart illustrating the planning method of the present invention.

[0066] Figure 3 This is a simplified schematic diagram of the vehicle's kinematics model. Detailed Implementation

[0067] The technical solution of the present invention will now be clearly and completely described in conjunction with the accompanying drawings.

[0068] Figure 1 This is a flowchart illustrating the implementation of the present invention. Figure 2 This is a flowchart illustrating the planning method of the present invention. Figure 1 , Figure 2 Therefore, this invention provides a risk-based intelligent vehicle lane-changing path planning method. The risk-based approach considers risk factors affecting the intelligent vehicle's direction, speed, and acceleration during operation to achieve autonomous lane-changing path planning. Specifically, the planning method of this invention includes the following steps:

[0069] Step 1, Establish the road boundary risk field E d_tq

[0070] The horizontal axis is set as the axis parallel to the horizon, and the positive direction is the front of the intelligent vehicle. The vertical axis is set as the axis perpendicular to the horizon in the ground plane, and the positive direction is the direction of rotating 90° counterclockwise from the front of the intelligent vehicle. The intersection of the two coordinate axes is the origin of the coordinate system. The road boundary and lane marking lines are parallel to the horizontal axis.

[0071] The risk distribution at the road boundary is assumed to exhibit a gradient descent trend, and the road boundary risk field E d_tq Described in power function form, its expression is:

[0072]

[0073] In the formula, μ is the road boundary field strength coefficient, and d tq Let d be the distance vector between the intelligent vehicle and the road boundary. tq d tq = (0, |y q -y t |-0.5(Lsinθ+wcosθ)), the vehicle side distance vector d tq The direction starts from the centroid of the intelligent vehicle and points to the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle. t Let y be the ordinate of the centroid of the intelligent vehicle. q Let L be the ordinate of the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle, w be the length of the intelligent vehicle, and θ be the heading angle of the intelligent vehicle.

[0074] Step 2, establish the lane marking risk field Eb_tm

[0075] The lane marking risk field E b_tm The expression is:

[0076]

[0077] In the formula, D m The risk coefficient for lane marking lines, d tm Let d be the distance vector between the intelligent vehicle and the lane marking line m. tm , Vehicle line distance vector d tm The direction starts from the centroid of the intelligent vehicle and points to the centroid of the lane marking line on the same horizontal coordinate as the intelligent vehicle. m The ordinate is the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle.

[0078] In this embodiment, the risk factor D of the lane marking line is... m The formula for calculation is:

[0079]

[0080] In the formula, T m Let d1 be the lane marking coefficient, and d1 be the lateral distance between the intelligent vehicle's centroid and the lane marking line m. d1 = |y t -y m |,y m Let w be the ordinate of the centroid of the lane marking line m, which shares the same x-coordinate as the centroid of the intelligent vehicle t. * This refers to the lane width.

[0081] Step 3, establish the obstacle risk field E z_ti

[0082] The obstacle risk field E z_ti The expression is:

[0083]

[0084] In the formula, λ is the first given constant, τ1 is the second given constant, and τ2 is the third given constant, all of which are greater than 0; b is the sequence number of the obstacle, b = 1, 2, meaning there are two types of obstacles, denoted as obstacle i. b Where obstacle i1 is an obstacle with negligible size, and obstacle i2 is an obstacle with non-negligible size; v ib For obstacle i b speed; d ti_b Let i be the distance vector between the intelligent vehicle and the obstacle. b The distance vector; For obstacle ib The velocity direction and the distance vector d from the obstacle ti_b The angle between the two points is denoted as the velocity-distance angle. M ib For equivalent mass, characterize the intelligent vehicle and obstacle i b The severity of the collision.

[0085] In this embodiment, the equivalent mass M ib The formula for calculation is:

[0086]

[0087] In the formula, m ib For obstacle i b Actual quality, T ib For obstacle i b The type coefficient, a is the velocity polynomial The number of terms, ξ a Given the fourth constant, η a Given the fifth constant, λ a It is the sixth given constant.

[0088] In this embodiment, the obstacle distance vector d ti_b The calculation is as follows:

[0089] For obstacle i1, the obstacle distance vector d ti_b The formula for calculation is:

[0090]

[0091] In the formula, t1 is the distance between obstacle i1 and the center of mass of the intelligent vehicle. x t Let x1 be the x-coordinate of the intelligent vehicle's center of mass, x1 be the x-coordinate of obstacle i1, and y1 be the y-coordinate of obstacle i1. Let d be the velocity direction of obstacle i1 and the distance vector d between the vehicle and the obstacle. ti_b The included angle.

[0092] For obstacle i2, the vehicle-obstacle distance vector d ti_b The formula for calculation is:

[0093]

[0094] In the formula, t2 is the distance between the obstacle i2 and the centroid of the intelligent vehicle t. x2 is the x-coordinate of the centroid of obstacle vehicle i2, y2 is the y-coordinate of the centroid of obstacle vehicle i2, β1 is the length of obstacle vehicle i2, β2 is the width of obstacle vehicle i2, α1 is an undetermined coefficient related to the length of obstacle i2, α2 is an undetermined coefficient related to the width of obstacle i2, v2 is the velocity of obstacle i2, and Ω is an undetermined coefficient related to the velocity of obstacle i2. t Let y be the x-coordinate of the centroid of the intelligent vehicle. t Let t be the ordinate of the centroid of the intelligent vehicle. Let d be the velocity direction of the center of mass of obstacle i2 and the distance vector d between the vehicle and the obstacle. ti_b The included angle.

[0095] In this embodiment, λ = 0.001, τ1 = 1, and τ2 = 0.05.

[0096] In this embodiment, obstacle i1 includes pedestrians. Obstacle i2 includes other vehicles traveling on the road.

[0097] Step 4, given the final objective function P of the lane-changing path planning.

[0098] Based on the road boundary risk field E d_tq Lane marking risk field E b_tm and obstacle risk field E z_ti By combining the vehicle kinematics model to establish a model predictive controller to realize lane-changing path planning, the final objective function P is obtained.

[0099] Step 4.1: Establish the vehicle kinematic model in the inertial coordinate system, its expression is:

[0100]

[0101] In the formula, Let be the derivative of the x-coordinate of the rear axle center of the intelligent vehicle. Let be the derivative of the longitudinal coordinate of the rear axle center of the intelligent vehicle. v is the derivative of the heading angle of the intelligent vehicle. tb Let denoted as the rear axle center velocity of the intelligent vehicle, ζ as the front wheel steering angle of the intelligent vehicle, and γ as the wheelbase of the intelligent vehicle.

[0102] Figure 3 This is a simplified schematic diagram of the vehicle's kinematics model.

[0103] Step 4.2, based on step 4.1, set the x-coordinate of the intelligent vehicle's centroid to x. t The centroid ordinate y of the intelligent vehicle t The intelligent vehicle uses its heading angle θ as a state variable and the front wheel steering angle ζ as a control variable to establish a model predictive controller, the expression of which is:

[0104]

[0105] In the formula, e is the current sampling time, e+1 is the next sampling time, and W is the sampling period.

[0106] Step 4.3, determine the final objective function P

[0107] E of the road boundary risk field d_tq As the objective function P1, the lane marking risk field E b_tm As the objective function P2, the obstacle risk field E z_ti As the objective function P3:

[0108]

[0109] In the formula, U1 is the prediction time domain at each sampling time, j is the step size at the current sampling time, j = 1, 2, ..., U1, E d_tq (j) represents the road boundary risk field E at the current sampling time and the j-th prediction step. d_tq E b_tm (j) represents the lane marking risk field E at the j-th prediction step time at the current sampling moment. b_tm E z_ti (j) represents the obstacle risk field E at the j-th prediction step time at the current sampling moment. z_ti .

[0110] The centroid of the intelligent vehicle is y-coordinated. t The error between the expected value and the target value is used as objective function P4, and the error between the intelligent vehicle's heading angle θ and the expected value is used as objective function P5, with the following expressions:

[0111]

[0112] In the formula, y t_w (j) represents the longitudinal position error of the intelligent vehicle at the j-th prediction step, θ w (j) represents the intelligent vehicle's heading angle error at the j-th prediction step, y t θ(j) represents the longitudinal position of the intelligent vehicle at the j-th prediction step, θ(j) represents the heading angle of the intelligent vehicle at the j-th prediction step, and y e (j) represents the expected longitudinal position of the intelligent vehicle at the j-th prediction step, θ e (j) represents the expected heading angle of the intelligent vehicle at the j-th prediction step.

[0113] Taking the front wheel steering angle ζ as the objective function P6, its expression is:

[0114]

[0115] In the formula, ζ(j) is the front wheel angle at the j-th prediction step, and U2 is the control time domain at each sampling time, satisfying U2≤U1;

[0116] Introducing the first weight coefficient ε1, the second weight coefficient ε2, the third weight coefficient ε3, the fourth weight coefficient ε4, the fifth weight coefficient ε5, and the sixth weight coefficient ε6, the final objective function P is expressed as follows:

[0117] P=ε1P1+ε2P2+ε3P3+ε4P4+ε5P5+ε6P6

[0118] In this embodiment, ε1 = 10, ε2 = 1000, ε3 = 1500, ε4 = 2, ε5 = 10, and ε6 = 150.

[0119] Step 4.4, combining the vehicle steering structure saturation threshold, describes the model predictive control optimization problem of path planning, and obtains the optimal front wheel steering angle ζ. best Its expression is:

[0120]

[0121] In the formula, ζ min The minimum value of the front wheel steering angle ζ, ζ max This represents the maximum value of the front wheel steering angle ζ.

[0122] Step 5: At each sampling time, the optimal front wheel steering angle ζ is obtained by solving for the objective function P and the vehicle steering structure saturation threshold. best Then return to step 4.2 to obtain the centroid x-coordinate of the intelligent vehicle at sampling time e+1 based on the model predictive controller. t The centroid ordinate y of the intelligent vehicle t The heading angle θ of the intelligent vehicle t is calculated; the above process is repeated at each sampling moment during the lane-changing process of the intelligent vehicle to obtain a real-time optimized lane-changing path.

Claims

1. A method for intelligent vehicle lane-changing path planning based on a risk field, characterized in that, Includes the following steps: Step 1, Establish the road boundary risk field E d_tq The horizontal axis is set as the axis parallel to the horizon, and the positive direction is the front of the intelligent vehicle. The vertical axis is set as the axis perpendicular to the horizon in the ground plane, and the positive direction is the direction of rotating 90° counterclockwise from the front of the intelligent vehicle. The intersection of the two coordinate axes is the origin of the coordinate system. The road boundary and lane marking lines are parallel to the horizontal axis. The risk distribution at the road boundary is assumed to exhibit a gradient descent trend, and the road boundary risk field E d_tq Described in power function form, its expression is: In the formula, μ is the road boundary field strength coefficient, and d tq Let d be the distance vector between the intelligent vehicle and the road boundary. tq d tq = (0, |y q -y t |-0.5(L sinθ+w cosθ)), the vehicle side distance vector d tq The direction starts from the centroid of the intelligent vehicle and points to the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle. t Let y be the ordinate of the centroid of the intelligent vehicle. q Let L be the ordinate of the centroid of the road boundary, which is on the same x-coordinate as the intelligent vehicle; w be the width of the intelligent vehicle; and θ be the heading angle of the intelligent vehicle. Step 2, establish the lane marking risk field E b_tm The lane marking risk field E b_tm The expression is: In the formula, D m The risk coefficient for lane marking lines, d tm Let d be the distance vector between the intelligent vehicle and the lane marking line m. tm , Vehicle line distance vector d tm The direction starts from the centroid of the intelligent vehicle and points to the centroid of the lane marking line on the same horizontal coordinate as the intelligent vehicle. m The ordinate of the centroid of the road boundary, which shares the same x-coordinate as the intelligent vehicle; Step 3, establish the obstacle risk field E z_ti The obstacle risk field E z_ti The expression is: In the formula, λ is the first given constant, τ1 is the second given constant, and τ2 is the third given constant, all of which are greater than 0; b is the sequence number of the obstacle, b = 1, 2, meaning there are two types of obstacles, denoted as obstacle i. b Where obstacle i1 is an obstacle with negligible size, and obstacle i2 is an obstacle with non-negligible size; v ib For obstacle i b speed; d ti_b Let i be the distance vector between the intelligent vehicle and the obstacle. b The distance vector; For obstacle i b The velocity direction and the distance vector d from the obstacle ti_b The angle between the two points is denoted as the velocity-distance angle. M ib For equivalent mass, characterize the intelligent vehicle and obstacle i b The severity of the collision; Step 4, given the final objective function P of the lane-changing path planning. Based on the road boundary risk field E d_tq Lane marking risk field E b_tm and obstacle risk field E z_ti By combining the vehicle kinematics model to establish a model predictive controller to realize lane-changing path planning, the final objective function P is obtained. Step 4.1: Establish the vehicle kinematic model in the inertial coordinate system, its expression is: In the formula, Let be the derivative of the x-coordinate of the rear axle center of the intelligent vehicle. Let be the derivative of the longitudinal coordinate of the rear axle center of the intelligent vehicle. v is the derivative of the heading angle of the intelligent vehicle. tb Let ζ be the rear axle center velocity of the intelligent vehicle, ζ be the front wheel steering angle of the intelligent vehicle denoted as front wheel steering angle ζ, and γ be the wheelbase of the intelligent vehicle. Step 4.2, based on step 4.1, set the x-coordinate of the intelligent vehicle's centroid to x. t The centroid ordinate y of the intelligent vehicle t The intelligent vehicle uses its heading angle θ as a state variable and the front wheel steering angle ζ as a control variable to establish a model predictive controller, the expression of which is: In the formula, e is the current sampling time, e+1 is the next sampling time, and W is the sampling period; Step 4.3, determine the final objective function P E of the road boundary risk field d_tq As the objective function P1, the lane marking risk field E b_tm As the objective function P2, the obstacle risk field E z_ti As the objective function P3: In the formula, U1 is the prediction time domain at each sampling time, j is the step size at the current sampling time, j = 1, 2, ..., U1, E d_tq (j) represents the road boundary risk field E at the current sampling time and the j-th prediction step. d_tq E b_tm (j) represents the lane marking risk field E at the j-th prediction step time at the current sampling moment. b_tm E z_ti (j) represents the obstacle risk field E at the j-th prediction step time at the current sampling moment. z_ti ; The centroid of the intelligent vehicle is y-coordinated. t The error between the expected value and the target value is used as objective function P4, and the error between the intelligent vehicle's heading angle θ and the expected value is used as objective function P5, with the following expressions: In the formula, y t_w (j) represents the longitudinal position error of the intelligent vehicle at the j-th prediction step, θ w (j) represents the intelligent vehicle's heading angle error at the j-th prediction step, y t θ(j) represents the longitudinal position of the intelligent vehicle at the j-th prediction step, θ(j) represents the heading angle of the intelligent vehicle at the j-th prediction step, and y e (j) represents the expected longitudinal position of the intelligent vehicle at the j-th prediction step, θ e (j) represents the expected heading angle of the intelligent vehicle at the j-th prediction step; Taking the front wheel steering angle ζ as the objective function P6, its expression is: In the formula, ζ(j) is the front wheel angle at the j-th prediction step, and U2 is the control time domain at each sampling time, satisfying U2≤U1; Introducing the first weight coefficient ε1, the second weight coefficient ε2, the third weight coefficient ε3, the fourth weight coefficient ε4, the fifth weight coefficient ε5, and the sixth weight coefficient ε6, the final objective function P is expressed as follows: P=ε1P1+ε2P2+ε3P3+ε4P4+ε5P5+ε6P6 Step 4.4, combining the vehicle steering structure saturation threshold, describes the model predictive control optimization problem of path planning, and obtains the optimal front wheel steering angle ζ. best Its expression is: In the formula, ζ min The minimum value of the front wheel steering angle ζ, ζ max This represents the maximum value of the front wheel steering angle ζ; Step 5: At each sampling time, the optimal front wheel steering angle ζ is obtained by solving for the objective function P and the vehicle steering structure saturation threshold. best Then return to step 4.2 to obtain the centroid x-coordinate of the intelligent vehicle at sampling time e+1 based on the model predictive controller. t The centroid ordinate y of the intelligent vehicle t The heading angle θ of intelligent vehicle i is obtained; the above process is repeated at each sampling moment during the lane-changing process of intelligent vehicle i to obtain a real-time optimized lane-changing path.

2. The intelligent vehicle lane-changing path planning method based on a risk field according to claim 1, characterized in that, The risk factor D of the lane marking lines m The formula for calculation is: In the formula, T m It is the lane marking coefficient, where d1 is the lateral distance between the intelligent vehicle's centroid and the lane marking line m, and d1 = |y t -y m |,y m The ordinate of the lane marking line m, which shares the same horizontal coordinate as the centroid of the intelligent vehicle t, is w. * It refers to the lane width.

3. The intelligent vehicle lane-changing path planning method based on a risk field according to claim 1, characterized in that, The equivalent mass M ib The formula for calculation is: In the formula, m ib For obstacle i b Actual quality, T ib For obstacle i b The type coefficient, a is the velocity polynomial The number of terms, ξ a Given the fourth constant, η a Given the fifth constant, λ a The sixth given constant; The obstacle distance vector d ti_b The calculation is as follows: For obstacle i1, the obstacle distance vector d ti_b The formula for calculation is: In the formula, t1 is the distance between obstacle i1 and the center of mass of the intelligent vehicle. x t Let x1 be the x-coordinate of the intelligent vehicle's center of mass, x1 be the x-coordinate of obstacle i1, and y1 be the y-coordinate of obstacle i1. Let d be the velocity direction of obstacle i1 and the distance vector d between the vehicle and the obstacle. ti_b The included angle; For obstacle i2, the vehicle-obstacle distance vector d ti_b The formula for calculation is: In the formula, t2 is the distance between the obstacle i2 and the centroid of the intelligent vehicle t. x2 is the x-coordinate of the centroid of obstacle vehicle i2, y2 is the y-coordinate of the centroid of obstacle vehicle i2, β1 is the length of obstacle vehicle i2, β2 is the width of obstacle vehicle i2, α1 is an undetermined coefficient related to the length of obstacle i2, α2 is an undetermined coefficient related to the width of obstacle i2, v2 is the velocity of obstacle i2, and Ω is an undetermined coefficient related to the velocity of obstacle i2. t Let y be the x-coordinate of the centroid of the intelligent vehicle. t Let t be the ordinate of the centroid of the intelligent vehicle. Let d be the velocity direction of the center of mass of obstacle B and the distance vector d between the obstacle and the vehicle. ti_b The included angle.

Images