Trajectory rapid resampling method, system and device based on unmanned motion constraint and medium

Abstract

The invention relates to the technical field of unmanned driving, in particular to a rapid track resampling method, system and device based on unmanned driving motion constraint and a medium. According to the method, vehicle kinematics is introduced, a virtual trolley is controlled to move along an expected track to generate a track meeting the kinematics, and the defect that a filtering method cannot give consideration to the kinematics is overcome; and meanwhile, in order to solve the problem that a control-based method cannot converge at the final target point of the trajectory, a forward filtering and backward filtering combined method is adopted, the lag problem caused by a pure filtering method is solved, and the advantage of high calculation speed of the filtering method is inherited to quickly perform trajectory resampling.

Description

Technical Field

[0001] The present invention relates to the field of unmanned driving technology, and in particular to a method, system, device and medium for rapid trajectory resampling based on unmanned driving motion constraints. Background Art

[0002] Autonomous vehicles typically follow a fixed trajectory during driving. Therefore, a series of discrete points must be recorded in advance as a reference trajectory to control the vehicle along this trajectory. The original recorded points are often noisy and sparse. Directly using such a trajectory for tracking may not meet driving comfort requirements and may even cause the vehicle to lose tracking due to sudden changes in curvature. Therefore, the original discrete points must be smoothed and resampled.

[0003] In the existing technology, optimization-based methods are usually used to process the original trajectory to meet kinematic constraints and improve trajectory smoothness. However, such methods have high computational complexity and high consumption of computing resources.

[0004] The Chinese invention patent, publication number "CN117367454A," titled "A method, device, apparatus, and storage medium for autonomous driving trajectory planning," which was published in Chinese patent literature, proposes a method for optimizing trajectory smoothing based on quadratic programming. This method can generate smooth trajectories and meet vehicle kinematic requirements, but the calculations are complex and time-consuming.

[0005] Similarly, the “A planning path reference line smoothing method based on curvature constraint pre-smoothing” disclosed in the Chinese patent literature, with publication number “CN115563440A” and application date: 2023.01.03, mentions a reference line smoothing method that is also a trajectory smoothing method based on quadratic programming optimization.

[0006] The Chinese invention patent with publication number "CN116400681A" and titled "A method and device for real-time fitting of the trajectory of unmanned logistics vehicles" mentions the use of a fifth-order polynomial fitting method to fit discrete points. This is slightly faster than the optimization-based method and can ensure curvature continuity, but it cannot guarantee that the minimum turning radius of the vehicle is met.

[0007] In addition to optimization-based methods and polynomial fitting-based methods, there are also filtering-based methods. Optimization-based methods are computationally complex and time-consuming. Fitting-based methods are more computationally time-consuming than optimization-based methods, but still quite time-consuming and require a two-stage approach: fitting followed by resampling. Filtering-based methods are computationally faster but struggle to meet kinematic constraints. Therefore, a method that can quickly resample trajectories while maintaining kinematic constraints is urgently needed. Summary of the Invention

[0008] In response to the problem that existing resampling methods are difficult to meet kinematic constraints, the present invention proposes a fast trajectory resampling method, system, equipment and medium based on unmanned driving motion constraints; the method introduces vehicle kinematics to control the virtual car to move along the desired trajectory to generate a trajectory that meets kinematics, thereby compensating for the shortcoming that the filtering method cannot take kinematics into account; at the same time, in order to solve the problem that the control-based method cannot converge at the final target point of the trajectory, a method combining forward filtering and backward filtering is used to solve the lag problem caused by the simple filtering method, and inherits the advantage of the fast calculation speed of the filtering method, so as to achieve rapid trajectory resampling while ensuring transportation.

[0009] The specific implementation contents of the present invention are as follows:

[0010] A fast trajectory resampling method based on unmanned driving motion constraints includes the following steps:

[0011] Step S1: convert the obtained original trajectory into a trajectory point list;

[0012] Step S2: Initialize the current point c according to the trajectory point list and assign it as the starting point. Find the point p closest to the current point c in the original trajectory, and start from point p to search for a point q outside the preview distance until point p is the last point of the original trajectory, completing the forward filtering.

[0013] Step S3: Initialize the current point c' and assign it as the end point, search for the point p' closest to the current point c' in the original trajectory, and start searching for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering;

[0014] Step S4: Repeat steps S2 to S3, and use the filtered trajectory as a new smooth trajectory.

[0015] In order to better implement the present invention, further, step S1 specifically includes the following steps:

[0016] Step S11: obtaining the original trajectory from the unmanned driving system;

[0017] Step S12: Convert the original trajectory into a trajectory point list Traj_Input stored in a set format, and define the first point of the trajectory as the starting point P start , the last point is defined as the end point P end , the trajectory point is defined as

[0018] In order to better implement the present invention, further, step S2 specifically includes the following steps:

[0019] Step S21: Initialize the current point c and assign it to P start , initialize an empty trajectory Traj_Output, initialize the filter coefficient;

[0020] Step S22: Find the point p closest to the current point c in the original trajectory, and starting from point p, sequentially search for the first point q beyond the preview distance preview_distance, and update the filter coefficient;

[0021] Step S23: Control the current point c to move in a direction approaching the point q by step length; the step length is the resampling interval;

[0022] Step S24: inserting the moved point c into the trajectory Traj_Output;

[0023] Step S25: Repeat steps S22 to S24 until point p is the last point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

[0024] In order to better implement the present invention, further, step S23 specifically includes the following steps:

[0025] Step S231: Calculate the position deviation dx and position deviation dy between the current point c and the point q;

[0026] Step S232: Calculate the straight-line distance d from the current point c to the point q based on the position deviation dx and the position deviation dy;

[0027] Step S233: Calculate the heading deviation α between the direction of the line connecting the current point c and the point q and the current point c;

[0028] Step S234: Calculate the heading deviation β of the line connecting point q, current point c, and point q based on the heading deviation α;

[0029] Step S235: Calculate a new curvature k based on the set control gain, straight-line distance d, heading deviation α, heading deviation β, and filter coefficient;

[0030] Step S236: Calculate the trajectory curvature radius r and the instantaneous center O of the trajectory based on the curvature k. x , new heading angle θ;

[0031] Step S237: Based on the instantaneous center O of the trajectory x , trajectory curvature radius r, heading angle θ, calculate the new coordinate value, and update the value of the current point c to (x, y, θ, k).

[0032] In order to better implement the present invention, further, step S3 specifically includes the following steps:

[0033] Step S31: Initialize the current point c' and assign it to P end , initialize an empty trajectory Traj_Output, initialize the filter coefficient;

[0034] Step S32: Search the original trajectory Traj_Input for the point p' closest to the current point c', and starting from point p', search in reverse order for the first point q' beyond the preview distance preview_distance, and update the filter coefficients;

[0035] Step S33: controlling the current point c' to move in a direction approaching the point q' by step length; the step length is the resampling interval;

[0036] Step S34: inserting the moved point c' into the trajectory Traj_Output;

[0037] Step S35: Repeat steps S32 to S34 until point p' is the first point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

[0038] Based on the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints, in order to better implement the present invention, a fast trajectory resampling system based on unmanned driving motion constraints is further proposed, which is used to execute the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints; the system includes an original trajectory input unit, a forward filtering unit, a backward filtering unit, and an iteration unit;

[0039] The original trajectory input unit is used to convert the obtained original trajectory into a trajectory point list;

[0040] The forward filtering unit is used to initialize the current point c according to the trajectory point list and assign it as the starting point, search for a point p closest to the current point c from the original trajectory, and start searching for a point q outside the preview distance from point p until point p becomes the last point of the original trajectory, completing the forward filtering;

[0041] The backward filtering unit is used to initialize the current point c' and assign it as the end point, search for a point p' closest to the current point c' from the original trajectory, and search for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering;

[0042] The iterative unit is used to repeat forward filtering and backward filtering, and use the filtered trajectory as a new smooth trajectory.

[0043] Based on the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints, in order to better implement the present invention, an electronic device is further proposed, including a memory and a processor; the memory stores a computer program; when the computer program is executed on the processor, the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints is implemented.

[0044] Based on the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints, in order to better implement the present invention, a computer-readable storage medium is further proposed, on which computer instructions are stored; when the computer instructions are executed on the above-mentioned electronic device, the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints is implemented.

[0045] The present invention has the following beneficial effects:

[0046] (1) The present invention combines forward filtering and backward filtering to quickly resample the trajectory while ensuring kinematic constraints, thus solving the lag problem caused by filtering.

[0047] (2) The present invention makes the calculation speed between the filtering method and the polynomial fitting method, thereby improving the calculation speed;

[0048] (3) The present invention introduces vehicle kinematics to control the virtual car to move toward the desired trajectory to generate a trajectory that satisfies kinematics, thereby compensating for the shortcoming that the filtering method cannot take kinematics into account; kinematics calculation is combined with filtering to solve the problem that simple filtering cannot satisfy geometric kinematics. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a schematic diagram of the original input trajectory provided by the present invention.

[0050] Figure 2 This is a schematic diagram of the heading deviation angle and instantaneous center of the circle provided by the present invention.

[0051] Figure 3 This is a schematic diagram of the kinematic geometry recursion provided by the present invention.

[0052] Figure 4 This is a schematic diagram of the smoothed trajectory provided by the present invention. DETAILED DESCRIPTION

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be understood that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments, and therefore should not be regarded as limiting the scope of protection. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technical personnel in this field without making creative work are within the scope of protection of the present invention.

[0054] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "disposed," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections, electrical connections; direct connections, indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0055] Example 1:

[0056] This embodiment proposes a fast trajectory resampling method based on unmanned driving motion constraints, which specifically includes the following steps:

[0057] A fast trajectory resampling method based on unmanned driving motion constraints includes the following steps:

[0058] Step S1: convert the obtained original trajectory into a trajectory point list;

[0059] The step S1 specifically includes the following steps:

[0060] Step S11: obtaining the original trajectory from the unmanned driving system;

[0061] Step S12: Convert the original trajectory into a trajectory point list Traj_Input stored in a set format, and define the first point of the trajectory as the starting point P start , the last point is defined as the end point P end , the trajectory point is defined as

[0062] Step S2: Initialize the current point c according to the trajectory point list and assign it as the starting point. Find the point p closest to the current point c in the original trajectory, and start from point p to search for a point q outside the preview distance until point p is the last point of the original trajectory, completing the forward filtering.

[0063] The step S2 specifically includes the following steps:

[0064] Step S21: Initialize the current point c and assign it to Pstart , initialize an empty trajectory Traj_Output, initialize the filter coefficient;

[0065] Step S22: Find the point p closest to the current point c in the original trajectory, and starting from point p, sequentially search for the first point q beyond the preview distance preview_distance, and update the filter coefficient;

[0066] Step S23: Control the current point c to move in a direction approaching the point q by step length; the step length is the resampling interval;

[0067] The step S23 specifically includes the following steps:

[0068] Step S231: Calculate the position deviation dx and position deviation dy between the current point c and the point q;

[0069] Step S232: Calculate the straight-line distance d from the current point c to the point q based on the position deviation dx and the position deviation dy;

[0070] Step S233: Calculate the heading deviation α between the direction of the line connecting the current point c and the point q and the current point c;

[0071] Step S234: Calculate the heading deviation β of the line connecting point q, current point c, and point q based on the heading deviation α;

[0072] Step S235: Calculate a new curvature k based on the set control gain, straight-line distance d, heading deviation α, heading deviation β, and filter coefficient;

[0073] Step S236: Calculate the trajectory curvature radius r and the instantaneous center O of the trajectory based on the curvature k. x , new heading angle θ;

[0074] Step S237: Based on the instantaneous center O of the trajectory x , trajectory curvature radius r, heading angle θ, calculate the new coordinate value, and update the value of the current point c to (x, y, θ, k).

[0075] Step S24: inserting the moved point c into the trajectory Traj_Output;

[0076] Step S25: Repeat steps S22 to S24 until point p is the last point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

[0077] Step S3: Initialize the current point c' and assign it as the end point, search for the point p' closest to the current point c' in the original trajectory, and start searching for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering;

[0078] The step S3 specifically includes the following steps:

[0079] Step S31: Initialize the current point c' and assign it to P end , initialize an empty trajectory Traj_Output, initialize the filter coefficient;

[0080] Step S32: Search the original trajectory Traj_Input for the point p' closest to the current point c', and starting from point p', search in reverse order for the first point q' beyond the preview distance preview_distance, and update the filter coefficients;

[0081] Step S33: controlling the current point c' to move in a direction approaching the point q' by step length; the step length is the resampling interval;

[0082] Step S34: inserting the moved point c' into the trajectory Traj_Output;

[0083] Step S35: Repeat steps S32 to S34 until point p' is the first point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

[0084] Step S4: Repeat steps S2 to S3, and use the filtered trajectory as a new smooth trajectory.

[0085] Working Principle: This embodiment introduces vehicle kinematics to control the virtual car to move toward the desired trajectory to generate a trajectory that satisfies kinematics, thereby compensating for the shortcoming of the filtering method that cannot take kinematics into account. At the same time, in order to solve the problem that the control-based method cannot converge at the final target point of the trajectory, a method combining forward filtering and backward filtering is used to solve the lag problem caused by the simple filtering method. It also inherits the advantage of the fast calculation speed of the filtering method, achieving rapid trajectory resampling while ensuring transportation.

[0086] Example 2:

[0087] This embodiment is based on the above embodiment 1. Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 As shown, a specific embodiment is described in detail.

[0088] This embodiment mainly generates a new smooth trajectory from the original trajectory by combining the ideas of filtering and control. Specifically, by introducing vehicle kinematics and controlling the virtual car to move toward the desired trajectory, a trajectory that satisfies kinematics is generated, compensating for the shortcomings of the filtering method that cannot take into account kinematics. At the same time, to solve the problem that the control-based method cannot converge at the final target point of the trajectory, this application proposes a method that combines forward filtering and backward filtering. This method simultaneously solves the lag problem caused by the simple filtering method and inherits the advantage of the filtering method's fast calculation speed.

[0089] Specifically, the trajectory smoothing method described in this application includes the following steps:

[0090] Step S1: Sample the original trajectory obtained from the unmanned driving system, such as a continuous list of path points obtained by the GNSS / RTK system for setting out a specific road path. The purpose is to convert the original trajectory into a list of trajectory points stored in a specified format, named Traj_Input. The specified trajectory points are composed of x-coordinates, y-coordinates, heading angles, and The curvature κ describes And save the first point of the trajectory as the starting point P start , the last point is the end point P end .

[0091] Step S2: Forward filtering.

[0092] Step S21: Initialize the current point c and assign it to P start , initialize an empty trajectory Traj_Output, initialize the filter coefficient λ = 1.0;

[0093] Step S22: Find the point p closest to the current point c in the original trajectory, and starting from point p, sequentially search for the first point q outside the set preview distance preview_distance, and update the filter coefficient λ = 0.9λ;

[0094] Step S23: Kinematic interpolation calculation: controlling the current point c to move closer to point q by a distance of step length step; where step is the resampling interval size;

[0095] Step S24: insert the new value of the current point, i.e., the point c after the movement, into the trajectory Traj_Output;

[0096] Step S25: Return to step S22 and execute in a loop until point p is the last point of the input trajectory, and overwrite Traj_Input with Traj_Output.

[0097] The kinematic interpolation in step S23 specifically includes the following steps:

[0098] Step S231: Calculate the position deviation between the current point c and the point q;

[0099] d x = x q - x c

[0100] d y = y q - y c

[0101] Step S232: Calculate the straight-line distance from the current point c to the point q;

[0102]

[0103] Step S233: Calculate the course deviation between the direction of the line connecting the current points c and q and the course of the current point c;

[0104]

[0105] where is the course angle of point c;

[0106] Step S234: Calculate the course deviation of the current point q from the direction of the line connecting the current points c and q;

[0107]

[0108] where is the course angle of point q;

[0109] Step S235: Set the control gains kp and kd. After testing, it works well when kp = 5 and kd = 2. Calculate the new curvature;

[0110]

[0111] Step S236: Calculate the radius

[0112] Limit the turning radius: Let the minimum physical turning radius of the vehicle be R. If |r| < R, then

[0113] The instantaneous center of the trajectory [[ID=6②]]

[0114] The new course angle where

[0115] Step S237: Calculate the new coordinates; x = O x + r × sin(θ) y = O y - r × cos(θ);

[0116] Step S238: Update the value of the current point c to (x, y, θ, κ).

[0117] Step S3: Backward filtering.

[0118] Step S31: Initialize the current point c' and assign it to P end Initialize an empty trajectory Traj_Output. Initialize the filter coefficient λ = 1.0;

[0119] Step S32: Search for the point p' closest to the current point c' in the trajectory Traj_Input, and starting from point p', search in reverse order for the first point q' beyond the preview distance preview_distance, and update λ = 0.9λ;

[0120] Step S33: kinematic interpolation calculation: controlling the current point c' to move towards the point q' by a distance of step length;

[0121] Step S34: insert the new value of the current point into the trajectory Traj_Output;

[0122] Step S35: Return to step S32 and execute the loop until point p' is the first point of the input trajectory. Overwrite Traj_Input with Traj_Output.

[0123] Working Principle: This embodiment is used to smooth and resample noisy, sparsely sampled trajectory points, such as GNSS measurement data, to obtain a smooth, kinematically constrained, dense trajectory.

[0124] This embodiment combines forward filtering and backward filtering to solve the lag problem caused by filtering, and uses kinematics calculation into the filtering module to solve the problem that simple filtering cannot meet geometric kinematics. The calculation speed is fast, and dense and smooth trajectory sampling is obtained in one go.

[0125] The rest of this embodiment is the same as that of the above-mentioned embodiment 1, and therefore will not be described in detail.

[0126] Example 3:

[0127] This embodiment, based on any one of the above embodiments 1 and 2, proposes a fast trajectory resampling system based on unmanned driving motion constraints, which is used to execute the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints; the system includes an original trajectory input unit, a forward filtering unit, a backward filtering unit, and an iteration unit;

[0128] The original trajectory input unit is used to convert the obtained original trajectory into a trajectory point list;

[0129] The forward filtering unit is used to initialize the current point c according to the trajectory point list and assign it as the starting point, search for a point p closest to the current point c from the original trajectory, and start searching for a point q outside the preview distance from point p until point p becomes the last point of the original trajectory, completing the forward filtering;

[0130] The backward filtering unit is used to initialize the current point c' and assign it as the end point, search for a point p' closest to the current point c' from the original trajectory, and search for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering;

[0131] The iterative unit is used to repeat forward filtering and backward filtering, and use the filtered trajectory as a new smooth trajectory.

[0132] This embodiment also proposes an electronic device, including a memory and a processor; the memory stores a computer program; when the computer program is executed on the processor, the above-mentioned unmanned driving motion constraint-based trajectory fast resampling method is implemented.

[0133] This embodiment further proposes a computer-readable storage medium having computer instructions stored thereon. When the computer instructions are executed on the above-mentioned electronic device, the above-mentioned fast trajectory resampling method based on unmanned driving motion constraints is implemented.

[0134] The rest of this embodiment is the same as any of the above-mentioned embodiments 1 and 2, and thus will not be described in detail.

[0135] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any simple modification or equivalent change made to the above embodiment based on the technical essence of the present invention shall fall within the scope of protection of the present invention.

Claims

1. A fast trajectory resampling method based on unmanned driving motion constraints, characterized in that: The specific steps include: Step S1: convert the obtained original trajectory into a trajectory point list; Step S2: Initialize the current point c according to the trajectory point list and assign it as the starting point. Find the point p closest to the current point c in the original trajectory. Then, start from point p and search for a point q outside the preview distance until point p is the last point of the original trajectory. This completes the forward filtering. Step S3: Initialize the current point c' and assign it as the end point, search for the point p' closest to the current point c' in the original trajectory, and start searching for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering; Step S4: Repeat steps S2 to S3, and use the filtered trajectory as a new smooth trajectory.

2. The fast trajectory resampling method based on unmanned driving motion constraints according to claim 1 is characterized in that: The step S1 specifically includes the following steps: Step S11: obtaining the original trajectory from the unmanned driving system; Step S12: Convert the original trajectory into a trajectory point list Traj_Input stored in a set format, and define the first point of the trajectory as the starting point P start , the last point is defined as the end point P end , the trajectory point is defined as (x,y,φ,κ).

3. The fast trajectory resampling method based on unmanned driving motion constraints according to claim 2 is characterized in that: The step S2 specifically includes the following steps: Step S21: Initialize the current point c and assign it to P start , initialize an empty trajectory Traj_Output, initialize the filter coefficient; Step S22: Find the point p closest to the current point c in the original trajectory, and starting from point p, sequentially search for the first point q beyond the preview distance preview_distance, and update the filter coefficient; Step S23: Control the current point c to move in a direction approaching the point q by step length; the step length is the resampling interval; Step S24: inserting the moved point c into the trajectory Traj_Output; Step S25: Repeat steps S22 to S24 until point p is the last point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

4. The fast trajectory resampling method based on unmanned driving motion constraints according to claim 3 is characterized in that: The step S23 specifically includes the following steps: Step S231: Calculate the position deviation dx and position deviation dy between the current point c and the point q; Step S232: Calculate the straight-line distance d from the current point c to the point q based on the position deviation dx and the position deviation dy; Step S233: Calculate the heading deviation α between the direction of the line connecting the current point c and the point q and the current point c; Step S234: Calculate the heading deviation β of the line connecting point q, current point c, and point q based on the heading deviation α; Step S235: Calculate a new curvature k based on the set control gain, straight-line distance d, heading deviation α, heading deviation β, and filter coefficient; Step S236: Calculate the trajectory curvature radius r and the instantaneous center O of the trajectory based on the curvature k. x , new heading angle θ; Step S237: Based on the instantaneous center O of the trajectory x , trajectory curvature radius r, heading angle θ, calculate the new coordinate value, and update the value of the current point c to (x, y, θ, k).

5. The fast trajectory resampling method based on unmanned driving motion constraints according to claim 2 is characterized in that: The step S3 specifically includes the following steps: Step S31: Initialize the current point c' and assign it to P end , initialize an empty trajectory Traj_Output, initialize the filter coefficient; Step S32: Search the original trajectory Traj_Input for the point p' closest to the current point c', and starting from point p', search in reverse order for the first point q' beyond the preview distance preview_distance, and update the filter coefficients; Step S33: controlling the current point c' to move in a direction approaching the point q' by step length; the step length is the resampling interval; Step S34: inserting the moved point c' into the trajectory Traj_Output; Step S35: Repeat steps S22 to S24 until point p' is the first point of the trajectory Traj_Output, and overwrite the trajectory point list Traj_Input with the trajectory Traj_Output.

6. A fast trajectory resampling system based on unmanned driving motion constraints, used to execute the fast trajectory resampling method based on unmanned driving motion constraints according to claim 1; characterized in that: It includes an original trajectory input unit, a forward filtering unit, a backward filtering unit, and an iteration unit; The original trajectory input unit is used to convert the obtained original trajectory into a trajectory point list; The forward filtering unit is used to initialize the current point c according to the trajectory point list and assign it as the starting point, search for a point p closest to the current point c from the original trajectory, and start searching for a point q outside the preview distance from point p until point p becomes the last point of the original trajectory, completing the forward filtering; The backward filtering unit is used to initialize the current point c' and assign it as the end point, search for a point p' closest to the current point c' from the original trajectory, and search for a point q' outside the preview distance from point p' until point p' becomes the first point of the original trajectory, completing the backward filtering; The iterative unit is used to repeat forward filtering and backward filtering, and use the filtered trajectory as a new smooth trajectory.

7. An electronic device, characterized in that: The invention comprises a memory and a processor; a computer program is stored on the memory; when the computer program is executed on the processor, the fast trajectory resampling method based on unmanned driving motion constraints as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions; when the computer instructions are executed on the electronic device according to claim 7, the fast trajectory resampling method based on unmanned driving motion constraints according to any one of claims 1 to 5 is implemented.

Images