Automatic driving three-dimensional track optimization method based on space-time passage corridor

Abstract

The invention relates to an automatic driving three-dimensional trajectory optimization method based on a space-time passage corridor. The method comprises the following steps: Q1, multi-source sensing fusion and space-time environment modeling; q2, initial track generation and multi-path sampling are carried out; q3, space-time corridor construction and safety space guarantee; q4, optimizing the track in the corridor; and Q5, outputting the final track and generating a motion control instruction. According to the method, the problem of insufficient safety certification of a traditional method in a complex dynamic scene is solved, and the problem of disjunction of decision and dynamics constraint caused by path and speed decoupling planning is solved by adopting a space-time joint optimization strategy.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to an optimization method for three-dimensional trajectories of autonomous driving based on spatiotemporal travel corridors. Background Technology

[0002] Autonomous driving trajectory planning is a core technology for ensuring the safe and efficient passage of vehicles in complex dynamic environments. 3D trajectory optimization, by introducing a time dimension, achieves simultaneous optimization of path geometry and velocity curves in the longitudinal-lateral-time (SLT) coordinate system, significantly improving the dynamic adaptability and scenario robustness of the trajectory. However, existing methods have obvious limitations: sampling-based methods, while capable of exploring the global space, suffer from high computational complexity and struggle to guarantee the continuous smoothness and spatiotemporal safety of the trajectory; graph search-based methods suffer from insufficient trajectory resolution and discontinuous higher-order derivatives due to discretization, making vehicle tracking difficult; and numerical optimization-based methods heavily rely on the quality of the initial solution, resulting in low efficiency when dealing with non-convex constraints introduced by dynamic obstacles. Furthermore, existing technologies often employ decoupled path and velocity planning, which can lead to module dispersion, inconsistencies between decision-making and dynamic constraints, and conservative performance in highly interactive scenarios such as forced lane changes and ramp merging, making it difficult to balance safety and traffic efficiency.

[0003] In the prior art, Chinese patent application (202311708459.2) discloses a lane matching method, apparatus, electronic device, and storage medium. The method includes: acquiring the current position and current direction of a target to be tracked; determining the current distance matching degree and current direction matching degree between the target to be tracked and multiple lanes based on the current position and current direction of the target to be tracked, and nearby lane information; wherein the lane information includes the lane position and lane direction of multiple lanes on the road; and determining the current lane matching result of the target to be tracked based on the current distance matching degree and current direction matching degree of the multiple lanes. However, the existing vehicle trajectory optimization method in this invention often relies on preset rules or simple models when dealing with complex scenarios such as intersections. It is difficult to effectively cope with dynamic traffic flow and complex interactive behaviors, resulting in insufficient flexibility in trajectory planning and susceptibility to the uncertainty of surrounding vehicle movement. This limits the accuracy and adaptability of the optimized trajectory and makes it difficult to meet the high-order requirements for safety and smoothness in practical applications.

[0004] In the prior art, Chinese patent application (202011119457.6) describes a method for optimizing vehicle 3D trajectory based on inverse perspective projection transformation and a vehicle-following model. This method comprises three parts: solving for vehicle 3D trajectory coordinates based on inverse perspective projection transformation, predicting vehicle motion coordinates based on a vehicle-following model, and optimizing and correcting vehicle 3D trajectory coordinates using a combination of inverse perspective projection transformation and the vehicle-following model. This invention combines the inverse perspective projection transformation method with a vehicle-following model from traffic simulation. However, existing vehicle 3D trajectory reconstruction methods have shortcomings due to their reliance on a single visual model. For example, inverse perspective projection transformation is easily affected by factors such as shooting angle and occlusion, resulting in limited accuracy of the initial trajectory coordinates and a lack of continuity in the temporal dimension. This makes it difficult to adaptively correct dynamic errors, thus affecting the smoothness, realism, and robustness of the reconstructed trajectory in complex scenarios. Summary of the Invention

[0005] In view of the above problems, this invention provides an optimization method for the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors. This method not only solves the problem of insufficient safety proof of traditional methods in complex dynamic scenarios, but also overcomes the problem of the disconnect between decision-making and dynamic constraints caused by the decoupling of path and speed planning by adopting a spatiotemporal joint optimization strategy.

[0006] To achieve the above and other related objectives, the present invention provides the following technical solution: An optimization method for the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors, the method comprising: Q1. Multi-source sensing fusion and spatiotemporal environment modeling; Q2. Initial trajectory generation and multipath sampling; Q3. Construction of the spacetime corridor and protection of safe space; Q4. Corridor trajectory optimization process; Q5. Output the final trajectory and generate motion control commands.

[0007] Furthermore, in step Q1, the multi-source sensing fusion and spatiotemporal environment modeling includes: Q11. Data preprocessing: Time synchronization and noise reduction processing of sensor data from LiDAR, cameras, millimeter-wave radar, GPS, and IMU; Q12. Perceptual Fusion: A deep learning fusion algorithm is used to generate a local point cloud map and extract the position, velocity, heading and semantic information of obstacles; Q13. Coordinate system transformation: Transform the global coordinates to the Frenet coordinate system, using the road centerline as the reference line, define the longitudinal displacement, lateral displacement and time dimension to form a three-dimensional spatiotemporal model.

[0008] Furthermore, the longitudinal displacement s and the lateral displacement l are, , , Where v(λ) is the velocity and θ(λ) is the heading angle.

[0009] Furthermore, the multi-source sensing fusion and spatiotemporal environment modeling also includes: Q14. Dynamic obstacle prediction: The future state of an obstacle is represented as a time series {(s)} i ,l i ,t i )}, lateral displacement s of dynamic obstacle obs (t) and longitudinal displacement l obs The calculation of (t) is as follows, s obs (t)=s0+v s t, l obs (t)=l0+l v t, Among them, v s Let l be the longitudinal velocity of the obstacle. v Let s0 be the lateral velocity of the obstacle, s0 be the initial value of the lateral displacement of the obstacle, and l0 be the initial value of the longitudinal displacement of the obstacle.

[0010] Furthermore, the initial trajectory generation and multipath sampling include: Q21. Trajectory Generation: Using the Lattice planner, multi-path sampling is performed in the Frenet coordinate system to generate candidate trajectories covering various driving behaviors. The trajectory is represented as a discrete path point sequence, with each point containing longitudinal, lateral, and timestamp coordinates (s, l, t). Q22. Trajectory Parameterization: The initial trajectory can be represented as a parameterized curve, with the following discrete form: , Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let k be the longitudinal displacement of the initial trajectory at time k.

[0011] Furthermore, in step Q3, the construction of the spatiotemporal corridor and the protection of the safe space include: Q31. Seed point generation: Discretize the initial trajectory into a spatiotemporal seed point sequence. , Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let be the longitudinal displacement of the initial trajectory at time k; Q32. Initial cube construction: For each pair of consecutive seed points (S i ,S i+1 Construct an axis-aligned initial cube. The boundary is defined as follows: ; Q33. Cube safe expansion algorithm: Expand the boundary of the cube step by step along three dimensions: S-axis (vertical), L-axis (horizontal), and T-axis (time axis), and perform collision detection at each step; Q34. Output the spacetime corridor: Spacetime corridor C = {C1, C2, C3, ..., C} n The output is a cube that defines a safe area. The optimization objective is to maximize the volume. .

[0012] Furthermore, in step Q33, the new boundary value b of the cube boundary new The calculation is b new =b old +dirΔstep, where b old is the initial boundary value of the cube boundary, dir is the direction, dir∈{-1,+1}, collision detection ensures that the cube has no intersection with the obstacle space, that is, CollisionFree(C) is not equal to the empty set.

[0013] Furthermore, in step Q4, the trajectory optimization process within the corridor includes: Q41. Trajectory parameterization: Using piecewise Bézier curves for each cubic interval [t] j-1 ,t j The trajectory is parameterized, with the longitudinal and lateral components represented by m-order Bézier curves. ,t∈[t j-1 ,t j ], Among them, α j =t j -t j-1 For time scale, It is a Bessel basis polynomial. For control points; Q42. Optimization problem modeling: The objective function is to minimize acceleration. , Where, ω s and ω l These are the weighting coefficients; Q43. Boundary Constraints: Trajectory points must lie within the corridor cube, i.e. ; Q44. Dynamic Constraints: Velocity acceleration , Among them, v max For the maximum speed of the vehicle, a max This is the vehicle's maximum acceleration.

[0014] Furthermore, the trajectory optimization process within the corridor also includes: Q45. Continuity constraint: At the boundaries of intervals, the position, velocity, and acceleration of the trajectory must be continuous, i.e. , k=0,1,2,...n.

[0015] Furthermore, in step Q5, the output of the final trajectory and generation of motion control commands includes: Q51. Obtain the optimal sequence of Bézier curve control points p from the optimizer. j i Substitute it into the equation of the Bézier curve, , The reconstructed trajectory curve is continuous and smooth. and , representing longitudinal and lateral movements respectively; Q52. Rigorous verification of the reconstructed trajectory includes: safety review, dynamics check, and continuity confirmation; Q53. Convert trajectory parameters into control commands: based on a continuous, smooth trajectory curve. Calculate the longitudinal velocity v(t) and longitudinal acceleration a(t). , , according to Calculate the transverse curvature h(t).

[0016] Mapped to the steering wheel angle, generating a time-stamp-aligned sequence of trajectory points (x k ,y k ,θ k ,v k ,a k ,t k ).

[0017] The present invention has the following positive effects: This invention provides a solution for trajectory planning of autonomous vehicles in complex dynamic scenarios that balances absolute safety, real-time computational efficiency, and ride comfort by combining spatiotemporal corridor construction with parameterized optimization of Bézier curves. The solution first utilizes spatiotemporal corridors to construct a convex passageway with strict safety guarantees in three-dimensional (longitudinal-lateral-temporal) space, ensuring that the vehicle trajectory does not collide with static or dynamic obstacles at any time, fundamentally overcoming the shortcomings of traditional planning methods that struggle to verify safety under continuous spatiotemporal constraints. Based on this, Bézier curves are used for parameterized trajectory optimization. Their convex hull property naturally satisfies corridor boundary constraints, and a high-order smooth trajectory is generated by minimizing objective functions such as jerk, significantly improving ride comfort. Simultaneously, a hierarchical optimization strategy—first generating a coarse trajectory as guidance, then performing refined optimization within the corridor—effectively reduces computational complexity, enabling the system to handle dynamic interactive scenarios in real time and avoiding the problem of spatial explosion. Ultimately, this method can generate safe, smooth, and dynamically feasible trajectories in various scenarios such as lane changes, intersections, and pedestrian interactions, significantly improving the intelligence, reliability, and environmental adaptability of autonomous driving systems. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram illustrating the construction of the spacetime corridor according to the present invention; Figure 3 This is a schematic diagram illustrating the corridor trajectory optimization of the present invention; Figure 4 This is a schematic diagram of the final trajectory generated in this invention; Figure 5 This is another schematic diagram illustrating the final trajectory generation of the present invention. Detailed Implementation

[0019] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0020] Example 1: As Figure 1 As shown, an optimization method for the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors is presented, the method comprising: Q1. Multi-source sensing fusion and spatiotemporal environment modeling; Q2. Initial trajectory generation and multipath sampling; Q3. Construction of the spacetime corridor and protection of safe space; Q4. Corridor trajectory optimization process; Q5. Output the final trajectory and generate motion control commands.

[0021] In this embodiment, in step Q1, the multi-source sensing fusion and spatiotemporal environment modeling includes: Q11. Data preprocessing: Time synchronization and noise reduction processing of sensor data from LiDAR, cameras, millimeter-wave radar, GPS, and IMU; Q12. Perceptual Fusion: A deep learning fusion algorithm is used to generate a local point cloud map and extract the position, velocity, heading and semantic information of obstacles; Q13. Coordinate system transformation: Transform the global coordinates to the Frenet coordinate system, using the road centerline as the reference line, define the longitudinal displacement, lateral displacement and time dimension to form a three-dimensional spatiotemporal model.

[0022] Furthermore, the longitudinal displacement s and the lateral displacement l are, , , Where v(λ) is the velocity and θ(λ) is the heading angle.

[0023] Furthermore, the multi-source sensing fusion and spatiotemporal environment modeling also includes: Q14. Dynamic obstacle prediction: The future state of an obstacle is represented as a time series {(s)} i ,l i ,t i )}, lateral displacement s of dynamic obstacle obs (t) and longitudinal displacement l obs The calculation of (t) is as follows, s obs (t)=s0+v s t, l obs (t)=l0+l v t, Among them, v s Let l be the longitudinal velocity of the obstacle. v Let s0 be the lateral velocity of the obstacle, s0 be the initial value of the lateral displacement of the obstacle, and l0 be the initial value of the longitudinal displacement of the obstacle.

[0024] In this embodiment, the initial trajectory generation and multipath sampling include: Q21. Trajectory Generation: Using the Lattice planner, multi-path sampling is performed in the Frenet coordinate system to generate candidate trajectories covering various driving behaviors. The trajectory is represented as a discrete path point sequence, with each point containing longitudinal, lateral, and timestamp coordinates (s, l, t). Q22. Trajectory Parameterization: The initial trajectory can be represented as a parameterized curve, with the following discrete form: , Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let k be the longitudinal displacement of the initial trajectory at time k.

[0025] In this embodiment, as Figure 2 As shown, in step Q3, the construction of the spatiotemporal corridor and the protection of the safe space include: Q31. Seed point generation: Discretize the initial trajectory into a spatiotemporal seed point sequence. , Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let be the longitudinal displacement of the initial trajectory at time k; Q32. Initial cube construction: For each pair of consecutive seed points (S i ,S i+1 Construct an axis-aligned initial cube. The boundary is defined as follows: ; Q33. Cube safe expansion algorithm: Expand the boundary of the cube step by step along three dimensions: S-axis (vertical), L-axis (horizontal), and T-axis (time axis), and perform collision detection at each step; Q34. Output the spacetime corridor: Spacetime corridor C = {C1, C2, C3, ..., C} n The output is a cube that defines a safe area. The optimization objective is to maximize the volume. .

[0026] In this embodiment, in step Q33, the new boundary value b of the cube boundary new The calculation is b new =b old +dirΔstep, where b old is the initial boundary value of the cube boundary, dir is the direction, dir∈{-1,+1}, collision detection ensures that the cube has no intersection with the obstacle space, that is, CollisionFree(C) is not equal to the empty set.

[0027] Example 2: Based on the optimization method of autonomous driving three-dimensional trajectory based on spatiotemporal corridor in Example 1, the present invention will be further explained and described below.

[0028] like Figure 1 As shown, an optimization method for the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors is presented, the method comprising: Q1. Multi-source sensing fusion and spatiotemporal environment modeling; Q2. Initial trajectory generation and multipath sampling; Q3. Construction of the spacetime corridor and protection of safe space; Q4. Corridor trajectory optimization process; Q5. Output the final trajectory and generate motion control commands.

[0029] In this embodiment, as Figure 3 As shown, in step Q4, the trajectory optimization process within the corridor includes: Q41. Trajectory parameterization: Using piecewise Bézier curves for each cubic interval [t] j-1 ,t j The trajectory is parameterized, with the longitudinal and lateral components represented by m-order Bézier curves. ,t∈[t j-1 ,t j ], Among them, α j =t j -t j-1 For time scale, It is a Bessel basis polynomial. For control points; Q42. Optimization problem modeling: The objective function is to minimize acceleration. , Where, ω s and ω l These are the weighting coefficients; Q43. Boundary Constraints: Trajectory points must lie within the corridor cube, i.e. ; Q44. Dynamic Constraints: Velocity acceleration , Among them, v max For the maximum speed of the vehicle, a max This is the vehicle's maximum acceleration.

[0030] In this embodiment, the trajectory optimization process within the corridor further includes: Q45. Continuity constraint: At the boundaries of intervals, the position, velocity, and acceleration of the trajectory must be continuous, i.e. , k=0,1,2,...n.

[0031] In this embodiment, as Figure 4 or Figure 5 As shown, in step Q5, the output of the final trajectory and generation of motion control commands includes: Q51. Obtain the optimal sequence of Bézier curve control points p from the optimizer. j i Substitute it into the equation of the Bézier curve, , The reconstructed trajectory curve is continuous and smooth. and , representing longitudinal and lateral movements respectively; Q52. Rigorous verification of the reconstructed trajectory includes: safety review, dynamics check, and continuity confirmation; Q53. Convert trajectory parameters into control commands: based on a continuous, smooth trajectory curve. Calculate the longitudinal velocity v(t) and longitudinal acceleration a(t). , , according to Calculate the transverse curvature h(t).

[0032] Mapped to the steering wheel angle, generating a time-stamp-aligned sequence of trajectory points (x k ,y k ,θ k ,v k ,a k ,t k ).

[0033] In this embodiment, the present invention provides an optimization system for an autonomous driving three-dimensional trajectory based on a spatiotemporal access corridor, including a computer device programmed or configured to execute the steps of the optimization method for an autonomous driving three-dimensional trajectory based on a spatiotemporal access corridor.

[0034] The present invention provides a computer-readable storage medium storing a computer program programmed or configured to perform the optimization method for autonomous driving three-dimensional trajectory based on spatiotemporal access corridors.

[0035] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0036] In summary, this invention not only solves the problem of insufficient security proof in complex dynamic scenarios by traditional methods, but also overcomes the problem of disconnect between decision-making and dynamic constraints caused by decoupled path and velocity planning by adopting a spatiotemporal joint optimization strategy.

[0037] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for optimizing the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors, characterized in that, The method includes: Q1. Multi-source sensing fusion and spatiotemporal environment modeling; Q2. Initial trajectory generation and multipath sampling; Q3. Construction of the spacetime corridor and protection of safe space; Q4. Corridor trajectory optimization process; Q5. Output the final trajectory and generate motion control commands.

2. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 1, characterized in that, In step Q1, the multi-source sensing fusion and spatiotemporal environment modeling includes: Q11. Data preprocessing: Time synchronization and noise reduction processing of sensor data from LiDAR, cameras, millimeter-wave radar, GPS, and IMU; Q12. Perceptual Fusion: A deep learning fusion algorithm is used to generate a local point cloud map and extract the position, velocity, heading and semantic information of obstacles; Q13. Coordinate system transformation: Transform the global coordinates to the Frenet coordinate system, using the road centerline as the reference line, define the longitudinal displacement, lateral displacement and time dimension to form a three-dimensional spatiotemporal model.

3. The method for optimizing the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors according to claim 2, characterized in that: The longitudinal displacement s and the lateral displacement l are, ， ， Where v(λ) is the velocity and θ(λ) is the heading angle.

4. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 2, characterized in that, The multi-source sensing fusion and spatiotemporal environment modeling also includes: Q14. Dynamic obstacle prediction: The future state of an obstacle is represented as a time series {(s)} i ,l i ,t i )}, lateral displacement s of dynamic obstacle obs (t) and longitudinal displacement l obs The calculation of (t) is as follows, s obs (t)=s0+v s t,l obs (t)=l0+l v t, Among them, v s Let l be the longitudinal velocity of the obstacle. v Let s0 be the lateral velocity of the obstacle, s0 be the initial value of the lateral displacement of the obstacle, and l0 be the initial value of the longitudinal displacement of the obstacle.

5. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 1, characterized in that, The initial trajectory generation and multipath sampling include: Q21. Trajectory Generation: Using the Lattice planner, multi-path sampling is performed in the Frenet coordinate system to generate candidate trajectories covering various driving behaviors. The trajectory is represented as a discrete path point sequence, with each point containing longitudinal, lateral, and timestamp coordinates (s, l, t). Q22. Trajectory Parameterization: The initial trajectory can be represented as a parameterized curve, with the following discrete form: ， Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let k be the longitudinal displacement of the initial trajectory at time k.

6. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal corridor according to claim 5, wherein in step Q3, the spatiotemporal corridor construction and safety space assurance include: Q31. Seed point generation: Discretize the initial trajectory into a spatiotemporal seed point sequence. ， Among them, t k =kΔt, where k is a positive integer and Δt is the sampling interval, s k Let l be the lateral displacement of the initial trajectory at time k. k Let be the longitudinal displacement of the initial trajectory at time k; Q32. Initial cube construction: For each pair of consecutive seed points (S i ,S i+1 Construct an axis-aligned initial cube. The boundary is defined as follows: ； Q33. Cube safe expansion algorithm: Expand the boundary of the cube step by step along three dimensions: S-axis (vertical), L-axis (horizontal), and T-axis (time axis), and perform collision detection at each step; Q34. Output the spacetime corridor: Spacetime corridor C = {C1, C2, C3, ..., C} n The output is a cube that defines a safe area. The optimization objective is to maximize the volume. 。 7. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 6, characterized in that, In step Q33, the new boundary value b of the cube boundary new The calculation is b new =b old +dirΔstep, where b old is the initial boundary value of the cube boundary, dir is the direction, dir∈{-1,+1}, collision detection ensures that the cube has no intersection with the obstacle space, that is, CollisionFree(C) is not equal to the empty set.

8. The method for optimizing the three-dimensional trajectory of autonomous driving based on spatiotemporal corridors according to claim 1, characterized in that, In step Q4, the trajectory optimization process within the corridor includes: Q41. Trajectory parameterization: Using piecewise Bézier curves for each cubic interval [t] j-1 ,t j The trajectory is parameterized, with the longitudinal and lateral components represented by m-order Bézier curves. ,t∈[t j-1 ,t j ], Among them, α j =t j -t j-1 For time scale, It is a Bessel basis polynomial. For control points; Q42. Optimization problem modeling: The objective function is to minimize acceleration. ， Where, ω s and ω l These are the weighting coefficients; Q43. Boundary Constraints: Trajectory points must lie within the corridor cube, i.e. ; Q44. Dynamic Constraints: Velocity acceleration , Among them, v max For the maximum speed of the vehicle, a max This is the vehicle's maximum acceleration.

9. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 8, characterized in that, The trajectory optimization process within the corridor also includes: Q45. Continuity constraint: At the boundaries of intervals, the position, velocity, and acceleration of the trajectory must be continuous, i.e. ， k=0,1,2,...n.

10. The method for optimizing the three-dimensional trajectory of autonomous driving based on a spatiotemporal passageway according to claim 8, characterized in that, In step Q5, the step of outputting the final trajectory and generating motion control commands includes: Q51. Obtain the optimal sequence of Bézier curve control points p from the optimizer. j i Substitute it into the equation of the Bézier curve, ， After reconstruction, a continuous and smooth trajectory curve is obtained. and , representing longitudinal and lateral movements respectively; Q52. Rigorous verification of the reconstructed trajectory includes: safety review, dynamics check, and continuity confirmation; Q53. Convert trajectory parameters into control commands: based on a continuous, smooth trajectory curve. Calculate the longitudinal velocity v(t) and longitudinal acceleration a(t). ， ， according to Calculate the transverse curvature h(t). ， Mapped to the steering wheel angle, generating a time-stamp-aligned sequence of trajectory points (x k ,y k ,θ k ,v k ,a k ,t k ).

Images