Intelligent automobile lane-changing obstacle-avoiding trajectory planning method based on case-based reasoning mechanism

Abstract

The invention relates to an intelligent automobile lane-changing obstacle-avoiding trajectory planning method based on a case-based reasoning mechanism. The method comprises the steps of case library establishment, case retrieval and reuse and case updating. The case library is established by the following steps: establishing a plurality of different cases based on a lane change scene, each case comprising a state quantity and an optimal decision quantity, the optimal decision quantity comprising a lane change obstacle avoidance trajectory T calculated based on a quintic polynomial lane change obstacle avoidance trajectory function and a safe distance S for executing lane change obstacle avoidance calculated based on a collision avoidance condition; the case retrieval and reuse comprises the steps that a case retrieval mode based on the distinction degree is adopted, the Euclidean distance serves as a distinction degree evaluation index, if the distinction degree of the current vehicle is smaller than or equal to a distinction degree threshold value, it is considered that similar cases exist in a case library, case reuse is executed, and if not, case update is executed. According to the method, the optimality and the real-time performance of the lane-changing obstacle-avoiding trajectory planning can be effectively considered.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent vehicle autonomous driving, and particularly relates to a method for intelligent vehicle lane-changing obstacle avoidance trajectory planning based on a case-based reasoning mechanism. Background Art

[0002] When a vehicle encounters an obstacle or a low-speed vehicle ahead, it needs to take lane-changing obstacle avoidance measures. During this process, the workload and pressure of human drivers will increase significantly, resulting in an increased risk of errors in operations and prone to traffic accidents. Through reasonable trajectory planning, intelligent vehicles can effectively perform autonomous lane-changing obstacle avoidance, thereby reducing the accident risk and improving traffic efficiency. Therefore, how to obtain a safe and reasonable lane-changing obstacle avoidance trajectory and ensure that the algorithm can meet the real-time requirement is crucial for improving the driving safety of intelligent vehicles.

[0003] Chinese Patent with application number 202311114503.7 proposes a method for intelligent vehicle obstacle avoidance path planning. By collecting the state information of the intelligent vehicle itself and surrounding vehicles through various on-vehicle sensors, it determines whether lane-changing is required, and considering the collision-free constraint conditions between the host vehicle and surrounding vehicles and its own dynamics constraints, a series of vehicle feasible path clusters are generated based on the B-spline algorithm, and the optimal path is selected through a multi-index comprehensive evaluation mechanism. However, this method does not study the planning time consumption.

[0004] The paper "Obstacle Avoidance and Formation Control of Multiple Unmanned Vehicles in Complex Environments Based on Artificial Potential Field Method" published in the 2nd issue of Volume 47 of Journal of University of Science and Technology Beijing in 2025 improves the artificial potential field method, solves the problems of collision between unmanned vehicles and obstacles and unreachability of the target point caused by excessive gravitational force in the traditional artificial potential field method, designs a new potential field function to generate a smoother obstacle avoidance path, and there is no complex constraint optimization problem during the obstacle avoidance process, with good real-time performance. However, it is difficult to achieve the optimality of the planned trajectory using this method.

[0005] In view of this, the present invention is specifically proposed. Summary of the Invention

[0006] The purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art, and propose a method for intelligent vehicle lane-changing obstacle avoidance trajectory planning based on a case-based reasoning (CBR) mechanism, laying a foundation for the safe and efficient driving of intelligent vehicles.

[0007] To achieve the above purpose, the present invention adopts the following technical solutions:

[0008] The present invention provides a method for intelligent vehicle lane-changing obstacle avoidance trajectory planning based on a case-based reasoning mechanism, which is characterized by including three parts: establishing a case base, case retrieval and reuse, and case update;

[0009] The establishment of the case library includes: based on the lane change scenario, multiple different cases are established. Each case includes state variables and optimal decision variables. The state variables all include the longitudinal vehicle speed of the host vehicle, the longitudinal vehicle speed of the obstacle vehicle, and the relative lateral displacement between the two vehicles. The optimal decision variables include the lane change obstacle avoidance trajectory T obtained based on the quintic polynomial lane change obstacle avoidance trajectory function and the safety distance S for executing lane change obstacle avoidance obtained based on the collision avoidance condition.

[0010] The case retrieval and reuse include: obtaining the state variables of the current vehicle, adopting a case retrieval method based on discrimination, using the Euclidean distance as the discrimination evaluation index, calculating the Euclidean distances between the state variables of the current vehicle and the state variables of each case in the case library respectively to obtain the corresponding discrimination degrees. When at least one of the discrimination degrees is less than or equal to the set discrimination degree threshold, it is considered that there are similar cases in the case library. Then, the safety distance S and the lane change obstacle avoidance trajectory T stored in the case corresponding to the minimum discrimination degree are output to execute case reuse. When all discrimination degrees are greater than the discrimination degree threshold, it is considered that there are no similar cases in the case library. Then, the lane change obstacle avoidance trajectory T and the safety distance S are determined through case update and the case is stored in the case library.

[0011] Further, the process of obtaining the lane change obstacle avoidance trajectory T is as follows:

[0012] The quintic polynomial lane change obstacle avoidance trajectory function is expressed as:

[0013]

[0014] where x(t) is the longitudinal coordinate of the lane change obstacle avoidance trajectory; y(t) is the lateral coordinate of the lane change obstacle avoidance trajectory; a i and b i (i = 1, 2,..., 5) are related constant coefficients; t represents time;

[0015] The host vehicle maintains a constant speed during the lane change obstacle avoidance process, and the longitudinal displacement W experienced during the lane change is the lane width. Then the following constraints hold:

[0016]

[0017] where v x,m is the longitudinal vehicle speed of the host vehicle; t0 is the start time of the host vehicle's lane change, and t0 = 0; t f is the end time of the lane change; D is the longitudinal displacement traveled during the vehicle's lane change, which can be calculated by the following formula:

[0018] D = v x,m t f (4)

[0019] Solving the simultaneous equations of equations (1) to (3) gives the coefficients in the fifth-degree polynomial lane-changing and obstacle-avoiding trajectory function, which are as follows:

[0020]

[0021] Considering the limitation of tire adhesion, the maximum lateral acceleration satisfies the following constraint:

[0022] a ymax = μg (7)

[0023] where a ymax is the maximum lateral acceleration; μ is the road surface adhesion coefficient; g is the acceleration due to gravity;

[0024] Based on equations (1), (5) and (6), the maximum lateral acceleration corresponding to the lane-changing and obstacle-avoiding trajectory T is obtained through extreme point analysis as:

[0025]

[0026] According to the linear two-degree-of-freedom vehicle model, the relationships between lateral acceleration, yaw rate, front wheel steering angle and longitudinal vehicle speed are:

[0027]

[0028]

[0029] where a y is the lateral acceleration; ω is the yaw rate; δ is the front wheel steering angle; L c is the wheelbase of the vehicle; K is the stability factor;

[0030] According to equations (8) to (10), the maximum yaw rate corresponding to the lane-changing and obstacle-avoiding trajectory T is:

[0031]

[0032] Taking into account the lateral acceleration, lane-changing time and yaw rate indexes comprehensively, and considering the differences in the orders of magnitude of the relevant indexes, normalization processing is carried out to establish the objective function as follows:

[0033]

[0034] Substituting equations (8) and (11) into equation (12) gives:

[0035]

[0036] In the formula, J(t f ) represents the objective function with t f as the optimization variable; ω maxRepresents the maximum yaw rate allowed during the lane-changing obstacle avoidance process, and its value is determined based on experience; t fmax Represents the longest lane-changing obstacle avoidance time, and its value is determined based on the statistical results of traffic data; t fmin Represents the minimum lane-changing obstacle avoidance time, and its value is determined according to the following formula;

[0037]

[0038] Solve equation (13) through the nonlinear programming method to obtain t f , and substitute it into equations (5) and (6) to obtain the correlation coefficients a i and b i , and then substitute a i and b i into equation (1) to obtain the lane-changing obstacle avoidance trajectory T.

[0039] Furthermore, the solution process of the safety distance S for performing lane-changing obstacle avoidance is as follows:

[0040] Assume that during the lane-changing process, the right front corner of the host vehicle collides with the left rear corner of the obstacle vehicle, and the collision point is c, and the corresponding time of the collision is t c , then there is:

[0041] 2y(t c ) - w m cosθ(t c ) = w o (15)

[0042] Among them, y(t c ) represents the lateral displacement of the host vehicle at the collision moment; w m is the width of the host vehicle; w o is the width of the obstacle vehicle; θ(t c ) is the heading angle of the host vehicle at the collision moment, which is calculated by the following formula:

[0043]

[0044] If the above collision occurs, the longitudinal displacements of the host vehicle and the obstacle vehicle satisfy the following relationship:

[0045]

[0046] Among them, x(t c ) and x o (t c ) respectively represent the longitudinal displacements of the host vehicle and the obstacle vehicle at time t c , and S0 is the longitudinal distance between the host vehicle and the obstacle vehicle at the initial moment;

[0047] Solving equations (1) and (15) - (17) simultaneously gives the initial distance at which the host vehicle collides with the obstacle vehicle as:

[0048]

[0049] where, v x,o is the longitudinal vehicle speed of the obstacle vehicle;

[0050] Therefore, the safety distance S for performing lane change to avoid obstacles is selected according to the following formula

[0051] S = S c + ΔS (19)

[0052] where, ΔS is a given constant representing the safety margin.

[0053] Furthermore, taking the Euclidean distance as the discrimination evaluation index, the calculation formula for the discrimination is:

[0054]

[0055] where, Sim i is the discrimination between the current vehicle state quantity and the state quantity of the i-th case; is the longitudinal vehicle speed of the current host vehicle, the longitudinal vehicle speed of the obstacle vehicle and is the relative lateral distance between the two vehicles, and are respectively the longitudinal vehicle speed of the host vehicle, the longitudinal vehicle speed of the obstacle vehicle and the relative lateral distance between the two vehicles stored in the i-th case in the case library.

[0056] Furthermore, the case update includes:

[0057] Using a cubic polynomial to fit the relationship between the lane change obstacle avoidance end time t f and the longitudinal vehicle speed v x,m of the host vehicle, obtaining fitting data by offline solving the optimal lane change obstacle avoidance end time corresponding to different longitudinal vehicle speeds of the host vehicle, and then obtaining the functional relationship between t f and v x,m as follows:

[0058]

[0059] where, A, B, C, and D are all constant coefficients determined by fitting;

[0060] Substituting the current longitudinal vehicle speed v x,m of the host vehicle and t f calculated according to equation (21) into equations (1), (5) and (6) can obtain the lane change obstacle avoidance trajectory T;

[0061] According to the current vehicle state variables, the safe distance S for lane-changing obstacle avoidance is determined by Equation (19).

[0062] Based on the current vehicle state variables, the lane-changing obstacle avoidance trajectory T, and the safe distance S for executing lane-changing obstacle avoidance, a new case is determined and stored in the case base to achieve case update.

[0063] Furthermore, the discrimination threshold is a constant given according to experience.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] An intelligent vehicle lane-changing obstacle avoidance trajectory planning method based on a case-based reasoning mechanism provided by the present invention can effectively balance the optimality and real-time performance of lane-changing obstacle avoidance trajectory planning. While ensuring the excellent performance of the planned trajectory, it can effectively reduce the planning time-consuming, laying a foundation for the safe and efficient driving of intelligent vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The accompanying drawings herein are incorporated into and constitute a part of this specification, and together with the specification are used to explain the principles of the present invention.

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0068] Figure 1 It is a schematic diagram of a typical lane-changing scenario of an intelligent vehicle;

[0069] Figure 2 It is a schematic diagram of the principle of the method proposed by the present invention;

[0070] Figure 3 It is a simulation result diagram of similar cases in the embodiment;

[0071] Figure 4 It is a simulation result diagram of the new case in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] Here, the exemplary embodiments will be described in detail. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the present invention. On the contrary, they are only examples consistent with some aspects of the present invention detailed in the appended claims.

[0073] Please refer to Figures 1 to 4, an embodiment of the present invention provides an intelligent vehicle lane-changing obstacle avoidance trajectory planning method based on a case-based reasoning mechanism, which includes three parts: establishing a case library, case retrieval and reuse, and case update;

[0074] The establishment of the case library includes: based on typical lane-changing scenarios, considering different longitudinal vehicle speeds of the host vehicle, longitudinal vehicle speeds of obstacle vehicles, and relative lateral distances between the two vehicles, establishing relevant cases and determining the state quantities of different cases, establishing multiple different cases, each case includes a state quantity and an optimal decision quantity, the state quantities all include the longitudinal vehicle speed of the host vehicle, the longitudinal vehicle speed of the obstacle vehicle, and the relative lateral distance between the two vehicles, and the optimal decision quantity includes a lane-changing obstacle avoidance trajectory T obtained based on a quintic polynomial lane-changing obstacle avoidance trajectory function and a safety distance S for executing lane-changing obstacle avoidance obtained based on collision avoidance conditions;

[0075] The case retrieval and reuse includes: obtaining the state quantity of the current vehicle, adopting a case retrieval method based on discrimination, using the Euclidean distance as the discrimination evaluation index, calculating the Euclidean distance between the state quantity of the current vehicle and the state quantities of each case in the case library respectively to obtain the corresponding discrimination, when at least one of the discriminations is less than or equal to the set discrimination threshold, it is considered that there are similar cases in the case library, and then output the stored safety distance S and lane-changing obstacle avoidance trajectory T in the case corresponding to the minimum discrimination (i.e., the optimal decision quantity) to execute case reuse; when all discriminations are greater than the discrimination threshold, it is considered that there are no similar cases in the case library, and then determine the lane-changing obstacle avoidance trajectory T and safety distance S through case update and store this case in the case library.

[0076] It should be noted that each time retrieval is performed, the state quantity of the current vehicle needs to be traversed with the state quantity of each case.

[0077] As Figure 1 shown is a schematic diagram of a typical lane-changing scenario of an intelligent vehicle. The host vehicle V m is driving straight at a longitudinal vehicle speed of v x,m on a one-way double-lane road, and its initial position is in the middle of the right lane, and the lane width is W. There is an obstacle vehicle V o on the driving route of the host vehicle, and its longitudinal vehicle speed is v x,o (v x,m >v x,o ). When detecting the obstacle vehicle ahead, the host vehicle needs to take lane-changing obstacle avoidance measures and effectively avoid obstacles through reasonable lane-changing obstacle avoidance trajectory planning.

[0078] The principle of the intelligent vehicle lane-changing obstacle avoidance trajectory planning method based on the case-based reasoning mechanism in the present invention is as Figure 2As shown in the figure. In this embodiment, it is assumed that the host vehicle always maintains a driving position in the middle of the right lane, and the origin of the coordinate system is defined at the midpoint of the front bumper of the host vehicle. There are three typical driving positions for the obstacle vehicle, namely, the position where the left side of the vehicle coincides with the left boundary of the lane, the middle position of the lane, and the position where the right side of the vehicle coincides with the right boundary of the road; considering that the longitudinal speeds of the host vehicle are 10 m / s, 15 m / s, 20 m / s, and 25 m / s respectively, and the longitudinal speeds of the obstacle vehicle are 5 m / s, 10 m / s, 15 m / s, and 20 m / s respectively. Since in the problem of lane-changing obstacle avoidance, the longitudinal speed of the host vehicle is greater than that of the obstacle vehicle, 30 sets of state variables can be formed based on the above settings.

[0079] Furthermore, based on the collision avoidance condition, the safe distance S for executing lane-changing obstacle avoidance is obtained, and the lane-changing obstacle avoidance trajectory T is solved based on the fifth-order polynomial lane-changing obstacle avoidance trajectory function, and finally the optimal decision-making quantity corresponding to each case is obtained.

[0080] The process of obtaining the lane-changing obstacle avoidance trajectory T in this embodiment is as follows:

[0081] The fifth-order polynomial lane-changing obstacle avoidance trajectory function is expressed as:

[0082]

[0083] where x(t) is the longitudinal coordinate of the lane-changing obstacle avoidance trajectory; y(t) is the lateral coordinate of the lane-changing obstacle avoidance trajectory; a i and b i (i = 1, 2,..., 5) are related constant coefficients; t represents time;

[0084] The host vehicle maintains a constant speed during the lane-changing obstacle avoidance process, and the longitudinal displacement W experienced during lane-changing is the lane width. Then the following constraints hold:

[0085]

[0086] where v x,m is the longitudinal speed of the host vehicle; t0 is the start time of the host vehicle's lane-changing. For the convenience of analysis, the start time of lane-changing obstacle avoidance is taken as the 0 moment, that is, t0 = 0; t f is the end time of lane-changing; D is the longitudinal displacement traveled during the vehicle's lane-changing process, which can be calculated by the following formula:

[0087] D = v x,m t f (4)

[0088] By solving the equations (1) to (3) simultaneously, the coefficients in the fifth-order polynomial lane-changing obstacle avoidance trajectory function are obtained as follows:

[0089]

[0090] Considering the limitation of tire adhesion, the maximum lateral acceleration satisfies the following constraint:

[0091] a ymax = μg (7)

[0092] where a ymax is the maximum lateral acceleration; μ is the road adhesion coefficient; g is the acceleration due to gravity;

[0093] Based on Eqs. (1), (5) and (6), the maximum lateral acceleration corresponding to the lane-changing obstacle avoidance trajectory T is obtained through extreme point analysis as:

[0094]

[0095] According to the linear two-degree-of-freedom vehicle model, the relationships among lateral acceleration, yaw rate, front wheel steering angle and longitudinal vehicle speed are:

[0096]

[0097] where a y is the lateral acceleration; ω is the yaw rate; δ is the front wheel steering angle; L c is the wheelbase of the vehicle; K is the stability factor;

[0098] According to Eqs. (8) - (10), the maximum yaw rate corresponding to the lane-changing obstacle avoidance trajectory T is:

[0099]

[0100] Taking into account the lateral acceleration, lane-changing time and yaw rate indexes comprehensively, and considering the differences in the orders of magnitude of the relevant indexes, normalization processing is carried out to establish the objective function as follows:

[0101]

[0102] Substituting Eqs. (8) and (11) into Eq. (12), we can get:

[0103]

[0104] In the formula, J(t f ) represents the objective function with t f as the optimization variable; ω max represents the maximum allowable yaw rate during the lane-changing obstacle avoidance process, and its value is determined according to experience. In this embodiment, ω max = 0.15; t fmax represents the longest lane-changing obstacle avoidance time, and its value is determined according to the traffic data statistical results. In this embodiment, t fmax = 9; t fmin represents the minimum lane-changing obstacle avoidance time, and its value is determined according to the following formula;

[0105]

[0106] Solve equation (13) by the nonlinear programming method to obtain t f , and substitute it into equations (5) and (6) to obtain the correlation coefficients a i and b i , and then substitute a i and b i into equation (1) to obtain the lane-changing obstacle avoidance trajectory T

[0107] Furthermore, considering the non-collision constraint of the vehicle, in order to avoid the vehicle colliding with the obstacle vehicle when driving along the planned trajectory, during lane-changing, the host vehicle V m and the obstacle vehicle V o should maintain a sufficient safety distance S

[0108] The solution process of the safety distance S for performing lane-changing obstacle avoidance is as follows

[0109] Assume that during the lane-changing process, the right front corner of the host vehicle just collides with the left rear corner of the obstacle vehicle, and the collision point is c, and the corresponding time of the collision is t c , then there is

[0110] 2y(t c ) - w m cosθ(t c ) = w o (15)

[0111] where y(t c ) represents the lateral displacement of the host vehicle at the collision time; w m is the width of the host vehicle; w o is the width of the obstacle vehicle; θ(t c ) is the heading angle of the host vehicle at the collision time, which is calculated by the following formula

[0112]

[0113] In addition, when the above collision situation just occurs, the longitudinal displacements of the host vehicle and the obstacle vehicle satisfy the following relationship

[0114]

[0115] where x(t c ) and x o (t c ) respectively represent the longitudinal displacements of the host vehicle and the obstacle vehicle at time t c , and S0 is the longitudinal distance between the host vehicle and the obstacle vehicle at the initial time

[0116] Solving equations (1) and (15) - (17) simultaneously gives the initial distance at which the host vehicle collides with the obstacle vehicle as:

[0117]

[0118] where v x,o is the longitudinal vehicle speed of the obstacle vehicle;

[0119] Therefore, the safety distance S for performing a lane change to avoid an obstacle is selected according to the following formula

[0120] S = S c + ΔS (19)

[0121] where ΔS is a given constant representing the safety margin, and its value can be determined according to experience. In this embodiment, ΔS = 1.

[0122] Furthermore, taking the Euclidean distance as the discrimination evaluation index, the calculation formula for the discrimination is:

[0123]

[0124] where Sim i is the discrimination between the current vehicle state quantity and the state quantity of the i-th case; is the longitudinal vehicle speed of the current host vehicle, the longitudinal vehicle speed of the obstacle vehicle, and is the relative lateral distance between the two vehicles, and are respectively the longitudinal vehicle speed of the host vehicle, the longitudinal vehicle speed of the obstacle vehicle, and the relative lateral distance between the two vehicles stored in the i-th case in the case library.

[0125] Furthermore, the case update includes:

[0126] Using a cubic polynomial to fit the relationship between the lane change obstacle avoidance end time t f and the longitudinal vehicle speed v x,m of the host vehicle, obtaining fitting data by offline solving the optimal lane change obstacle avoidance end time corresponding to different longitudinal vehicle speeds of the host vehicle, and then obtaining the functional relationship between t f and v x,m as follows:

[0127]

[0128] where A, B, C, and D are all constant coefficients determined by fitting;

[0129] Substituting the current longitudinal vehicle speed v x,m of the host vehicle and the t f calculated according to equation (21) into equations (1), (5), and (6) can obtain the lane change obstacle avoidance trajectory T;

[0130] According to the current vehicle state variables, determine the safe distance S for lane-changing obstacle avoidance through Equation (19).

[0131] Based on the current vehicle state variables, the lane-changing obstacle avoidance trajectory T, and the safe distance S for executing lane-changing obstacle avoidance, determine a new case and store it in the case base to achieve case update.

[0132] Specifically, the discrimination threshold is a constant given according to experience, and it is taken as 2 in this embodiment.

[0133] To prove the effectiveness and real-time performance of the present invention, the following experiments were conducted:

[0134] Use PreScan to build an intelligent vehicle lane-changing obstacle avoidance scenario, and use Matlab / Simulink for algorithm development, so as to establish a joint simulation experiment platform of PreScan and Matlab / Simulink. On this basis, verify the proposed method by selecting two scenarios of similar cases and new cases in the case base respectively, and select the method based on online optimization of the fifth-degree polynomial as the comparison method. The relevant simulations are carried out on the same scientific research computer.

[0135] The scenario setting of similar cases (when there are similar cases in the case base for the current scenario) is as follows: The initial position of the host vehicle is x m = 0 m, y m = 0 m, the longitudinal vehicle speed is v x,m = 25 m / s, the initial position of the obstacle vehicle is x o = 32 m, y o = 0 m, the longitudinal vehicle speed is v x,o = 15 m / s, the discrimination threshold Sim t = 2, the safety margin ΔS = 1 m. The simulation results are as Figure 3 shown. It can be seen that both the method of the present invention and the comparison method can achieve lane-changing obstacle avoidance trajectory planning, and the planned trajectories coincide, indicating that the performance of the trajectories planned by the two methods is the same. However, since the method of the present invention can directly call the optimal trajectory stored in the case base, the planning time is shorter, only 1.711×10 -4 s; while the comparison method requires 0.0014 s due to online optimization.

[0136] The scenario setting of new cases (when there are no similar cases in the case base for the current scenario) is as follows: The initial position of the host vehicle is x m = 0 m, y m = 0 m, the longitudinal vehicle speed of the host vehicle is v x,m = 23 m / s, the initial position of the obstacle vehicle is x o = 39 m, y o = 0.325 m, the longitudinal vehicle speed of the obstacle vehicle is v x,o= 11 m / s, discrimination threshold Sim t = 2, safety margin ΔS = 1 m. The simulation results are as follows Figure 4 shown. It can be seen that when encountering a new case, the method of the present invention can correct the decision-making quantity in real time according to the state quantity, and plan a lane-changing obstacle avoidance trajectory, and its trajectory performance is close to that of the comparative method. However, in terms of the planning time, the method of the present invention is only 0.0011 s, while the comparative method requires 0.0029 s.

[0137] It should be noted that the comparative method is a lane-changing obstacle avoidance trajectory planning method based on online optimization of a fifth-degree polynomial.

[0138] In summary, the method of the present invention can effectively reduce the planning time while ensuring the excellent performance of the planned trajectory. Compared with the comparative method, the planning time is reduced by 87.78% in the case of similar scenarios and 62.07% in the case of new scenarios. It should be noted that the on-vehicle computing platform installed in a real intelligent vehicle generally does not have as high a configuration as a scientific research computer, and the advantage of the method of the present invention in terms of the planning time will be further highlighted.

[0139] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention.

[0140] It should be understood that the present invention is not limited to the content already described above, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A method for intelligent vehicle lane change and obstacle avoidance trajectory planning based on case-based reasoning mechanism, characterized in that: It includes three parts: establishing case library, case retrieval and reuse, and case update; The establishment of the case library includes: based on the lane changing scenario, establishing a plurality of different cases, each of the cases includes a state quantity and an optimal decision quantity, the state quantities include the longitudinal speed of the main vehicle, the longitudinal speed of the obstacle vehicle and the relative lateral displacement of the two vehicles, the optimal decision quantity includes a lane changing obstacle avoidance trajectory T obtained based on a quintic polynomial lane changing obstacle avoidance trajectory function and a safe distance S for executing lane changing obstacle avoidance obtained based on a collision avoidance condition; The case retrieval and reuse includes: obtaining the state quantity of the current vehicle, adopting a case retrieval method based on the degree of distinction, taking the Euclidean distance as the degree of distinction evaluation index, respectively calculating the Euclidean distance between the state quantity of the current vehicle and the state quantity of each case in the case library, and obtaining the corresponding degree of distinction; when at least one of the degrees of distinction is less than or equal to the set degree of distinction threshold, it is considered that there are similar cases in the case library, and then the safety distance S and the lane change obstacle avoidance trajectory T stored in the case corresponding to the minimum degree of distinction are output to perform case reuse; when all the degrees of distinction are greater than the degree of distinction threshold, it is considered that there are no similar cases in the case library, and then the lane change obstacle avoidance trajectory T and the safety distance S are determined by case update, and the case is stored in the case library.

2. The intelligent vehicle lane change obstacle avoidance trajectory planning method based on case-based reasoning mechanism according to claim 1 is characterized in that: The process of obtaining the lane change obstacle avoidance trajectory T is as follows: The quintic polynomial lane change obstacle avoidance trajectory function is expressed as: Among them, x(t) is the longitudinal coordinate of the lane change obstacle avoidance trajectory; y(t) is the lateral coordinate of the lane change obstacle avoidance trajectory; a i and b i (i=1,2,…,5) is the related constant coefficient; t represents time; The vehicle maintains a constant speed during the lane change and obstacle avoidance process, and the longitudinal displacement W experienced during the lane change is the lane width, then the following constraints hold: Among them, v x,m is the longitudinal speed of the main vehicle; t0 is the time when the main vehicle starts to change lanes, t0=0; t f is the lane change end time; D is the longitudinal displacement of the vehicle during the lane change process, which can be calculated by the following formula: D=v x,m t f (4) The coefficients of the fifth-order polynomial lane-changing obstacle avoidance trajectory function are obtained by solving equations (1) to (3), as follows: Considering the limitation of tire adhesion, the maximum lateral acceleration satisfies the following constraints: a ymax =μg (7) Among them, a ymax is the maximum lateral acceleration; μ is the road adhesion coefficient; g is the gravitational acceleration; Based on equations (1), (5) and (6), the maximum lateral acceleration corresponding to the lane change obstacle avoidance trajectory T is obtained through extreme point analysis: According to the linear two-degree-of-freedom vehicle model, the relationship between lateral acceleration, yaw rate, front wheel angle and longitudinal speed is: Among them, a y is the lateral acceleration; ω is the yaw rate; δ is the front wheel turning angle; L c is the vehicle wheelbase; K is the stability factor; According to equations (8) to (10), the maximum yaw angular velocity corresponding to the lane change obstacle avoidance trajectory T is obtained as: Taking into account the lateral acceleration, lane change time and yaw rate indicators, and considering the differences in the order of magnitude of the relevant indicators, normalization is performed and the objective function is established as follows: Substituting equations (8) and (11) into equation (12), we can obtain: In the formula, J(t f ) indicates that t f is the objective function of the optimized variable; ω max Indicates the maximum yaw rate allowed during lane change and obstacle avoidance; t fmax Indicates the longest lane change obstacle avoidance time; t fmin It represents the minimum lane change obstacle avoidance time, and its value is determined according to the following formula; Solving equation (13) by nonlinear programming method yields t f , and substitute it into equations (5) and (6) to obtain the correlation coefficient a i and b i , and then a i and b i Substituting into equation (1) we can get the lane change obstacle avoidance trajectory T.

3. The intelligent vehicle lane change obstacle avoidance trajectory planning method based on case-based reasoning mechanism according to claim 2 is characterized in that: The process of solving the safe distance S for lane change and obstacle avoidance is as follows: Assume that during the lane change process, the right front corner of the main vehicle collides with the left rear corner of the obstacle vehicle, and the collision point is c, and the corresponding time when the collision occurs is t c , then: 2y(t c )-w m cosθ(t c )=w o (15) Among them, y(t c ) represents the lateral displacement of the main vehicle at the time of collision; w m w is the width of the main vehicle; o is the width of the obstacle vehicle; θ(t c ) is the heading angle of the main vehicle at the time of collision, which is calculated by the following formula: If the above collision occurs, the longitudinal displacement of the main vehicle and the obstacle vehicle satisfies the following relationship: Among them, x(t c ) and x o (t c ) represent t c The longitudinal displacement of the main vehicle and the obstacle vehicle at the moment, S0 is the longitudinal distance between the main vehicle and the obstacle vehicle at the initial moment; The initial distance between the main vehicle and the obstacle vehicle when they collide is obtained by solving equations (1) and (15) to (17): Among them, v x,o is the longitudinal speed of the obstacle vehicle; Therefore, the safe distance S for lane change and obstacle avoidance is selected according to the following formula: S=S c +ΔS (19) Among them, ΔS is a given constant, which represents the safety margin.

4. The intelligent vehicle lane change obstacle avoidance trajectory planning method based on case-based reasoning mechanism according to claim 3 is characterized in that: Taking Euclidean distance as the discrimination evaluation index, the calculation formula of the discrimination is: Among them, Sim i is the difference between the current vehicle state and the i-th case state; is the current longitudinal speed of the main vehicle, The longitudinal speed of the obstacle vehicle is the relative lateral distance between the two vehicles, and They are respectively the longitudinal speed of the main vehicle, the longitudinal speed of the obstacle vehicle and the relative lateral distance between the two vehicles stored in the i-th case in the case library.

5. The intelligent vehicle lane change obstacle avoidance trajectory planning method based on case-based reasoning mechanism according to claim 4 is characterized in that: The case updates include: Use cubic polynomial to calculate the end time t of lane change and obstacle avoidance f The longitudinal speed v of the main vehicle x,m The relationship between t and t is fitted, and the optimal lane change obstacle avoidance end time corresponding to different longitudinal speeds of the main vehicle is solved offline to obtain the fitting data, and then t f With v x,m The functional relationship is as follows: Among them, A, B, C, and D are constant coefficients determined by fitting; The current longitudinal speed of the main vehicle v x,m and t calculated according to formula (21) f Substituting into equations (1), (5) and (6), we can obtain the lane change obstacle avoidance trajectory T; According to the current vehicle state, the safe distance S for lane change and obstacle avoidance is determined by formula (19); Based on the current vehicle state, lane change obstacle avoidance trajectory T, and safe distance S for lane change obstacle avoidance, a new case is determined and stored in the case library to achieve case update.

6. The intelligent vehicle lane change obstacle avoidance trajectory planning method based on case-based reasoning mechanism according to claim 4 is characterized in that: The distinction threshold is a constant given based on experience.

Images