A path planning method and system for a quadrotor unmanned aerial vehicle

Abstract

The application discloses a path planning method and system of a four-rotor unmanned aerial vehicle, relates to the technical field of autonomous flight and path planning of unmanned aerial vehicles, and comprises the following steps: acquiring sensing data of a target region, establishing a multi-resolution map, and generating an initial path based on the multi-resolution map; correcting the initial path and constructing a safety corridor; and outputting a trajectory by taking the safety corridor as a constraint condition; and the application solves the problems of path continuity and dynamic obstacle avoidance by means of the multi-resolution map and the safety corridor.

Description

Technical Field

[0001] This invention relates to the field of autonomous flight and path planning technology for unmanned aerial vehicles (UAVs), and more specifically, to a path planning method and system for a quadcopter UAV. Background Technology

[0002] With the widespread application of drones in aerial photography, logistics delivery, inspection, disaster relief, and other fields, quadcopter drones have become a research hotspot due to their maneuverability, simple structure, and strong hovering ability. However, autonomous flight in complex environments still faces many technical challenges, especially in path planning. Existing technologies suffer from poor path continuity, insufficient obstacle avoidance capabilities, inaccurate mission timing control, and limited real-time adaptability.

[0003] For example, the invention patent with publication number CN107764255A discloses a drone path planning method. By identifying the edge of the building's roof and planning the flight route based on the edge, it can ensure that the drone flies over the building's roof as much as possible, so that even if objects or the drone itself fall, it will not cause harm to pedestrians. When there is no building in the area, the drone will use image recognition to ensure that there are no moving objects before continuing to move, thus improving safety.

[0004] For example, the invention patent with publication number CN103809597B discloses a flight path planning method for a drone and a drone. The method includes the following steps: acquiring depth information of the drone's flight environment and generating a two-dimensional grid map of the flight environment based on the depth information, the flight environment including obstacles; constructing a potential function for each grid and corresponding obstacle based on the position of each grid and obstacle in the two-dimensional grid map; obtaining a weighted graph of the grids in the two-dimensional grid map based on the drone's position and the potential function; and determining the flight path based on the weighted graph. According to this embodiment, the method constructs a potential function using a two-dimensional grid map and determines the flight path using a weighted graph obtained from the potential function, improving computational efficiency and exhibiting good scalability.

[0005] The above-disclosed technical solutions have at least the following technical problems:

[0006] In traditional techniques, most path planning methods rely solely on discrete sampling or grid-based search, often resulting in polygonal paths with sharp turns, leading to uneven flight, increased energy consumption, and higher dynamic burden. Furthermore, traditional methods often separate global path planning from local obstacle avoidance, lacking a unified optimization mechanism and prone to local optima or globally infeasible paths. To address these issues, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a path planning method and system for a quadcopter unmanned aerial vehicle (UAV), which solves the problems of path continuity and dynamic obstacle avoidance through multi-resolution maps and safety corridors.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A path planning method for a quadcopter unmanned aerial vehicle (UAV) includes: acquiring perception data of a target area, establishing a multi-resolution map, and generating an initial path based on the multi-resolution map; correcting the initial path and constructing a safe corridor; and outputting the trajectory with the safe corridor as a constraint.

[0010] In a preferred embodiment, the steps of acquiring perception data of the target area, establishing a multi-resolution map, and generating an initial path based on the multi-resolution map are as follows: Acquiring perception data of the target area; establishing a multi-resolution map based on the perception data, the multi-resolution map including a global map and a local map; determining the starting and target positions of the UAV in the global map according to task requirements, and identifying key access directions based on the global topology; constraining the UAV's search space based on the global and local maps, combined with the UAV's starting and target positions, and generating initial path candidate solutions within the constrained search space; performing segment-by-segment splicing processing on the initial path candidate solutions using the principle of curvature continuity to form a geometric path that meets continuity requirements; allocating expected arrival time windows at key nodes based on the geometric path; and transforming the geometric path into an initial path with temporal attributes by assigning time parameterization constraints to the key nodes.

[0011] In a preferred embodiment, the search space of the UAV is constrained based on the global map and local map, combined with the starting and target positions of the UAV, and initial path candidate solutions are generated within the constrained search space. Specifically, the following steps are taken: In the global map, the task area of ​​the UAV is determined using the starting and target positions as boundaries. Combined with obstacle distribution information in the local map, the task area is cropped to generate a constrained search space. The search space is then divided into dense obstacle areas and open areas. When a dense obstacle area exists in the search space, based on the spatial distribution information of the obstacles, an importance sampling mechanism is used to increase the node sampling probability in narrow passage areas, resulting in a dense area sampling point set. When an open area exists in the search space, a uniform sampling strategy is adopted to evenly distribute sampling nodes within the open area, resulting in an open area sampling point set. The dense area sampling point set and the open area sampling point set are merged to construct a unified candidate node set. Using this candidate node set as input, a path search algorithm generates several coarse path candidate solutions.

[0012] In a preferred embodiment, the step of obtaining the expected arrival time window is as follows: Based on the maximum and minimum speeds of the UAV and the geometric length of the path segment, the theoretical shortest and longest flight times for each path segment are calculated to obtain the initial time interval for that path segment; the future trajectory of dynamic obstacles is estimated using a kinematic prediction model; if an obstacle is predicted to occupy a relevant path segment near a critical node, the time interval conflicting with the obstacle is eliminated within the initial time interval, thereby narrowing the feasible time range and forming a time interval constrained by the obstacle; the time interval constrained by the obstacle is corrected according to the local environmental complexity; and the corrected time interval is used as the expected arrival time window for the critical node.

[0013] In a preferred embodiment, the initial path correction is specifically as follows: traversing the node sequence of the initial path, when the directional change between two adjacent path segments is less than a preset geometric threshold and there is no conflict in the corresponding time windows, deleting redundant intermediate nodes, and simultaneously merging and inheriting the time windows of the deleted nodes with the preceding and following nodes; detecting key points in the path where the turning angle exceeds the threshold, replacing the original polyline with a smooth curve segment with continuous curvature to obtain an initial local curve segment; recalculating the shortest and longest flight times of the segment based on the geometric length of the replaced curve segment, comparing them with the corresponding time windows, and updating the node's time window; acquiring the UAV's dynamic data, and using the dynamic parameters as constraints to establish a feasibility boundary for the path segment; and so on. The initial local curve segment is used as the initial solution. A multi-objective optimization function is constructed based on the feasibility boundary of the path segment. This function is then transformed into a convex optimization problem, and the control points of the path segment are optimized using quadratic programming to obtain the optimized local correction curve. After obtaining the local correction curve, a bidirectional check is performed on the time window of the entire path. When a conflict is detected, the velocity profile is adjusted first, increasing or decreasing the local velocity if the boundary conditions allow. If the conflict cannot be resolved, insertion waiting or backtracking to the candidate path set is used. The local correction curves are then concatenated in the path order to obtain the corrected initial path. The corrected initial path is projected into the time-space domain, and collision detection of static and dynamic obstacles is performed. If the detection passes, the output is the final path.

[0014] In a preferred embodiment, the construction of the safe corridor is specifically as follows: the initial path is divided into multiple path segments according to key nodes, each path segment corresponding to a clear start and end point and an expected arrival time window; for each path segment, in conjunction with the obstacle distribution in the local map, a candidate passable area is generated around the path segment using a multi-scale expansion method; if the candidate area conflicts with the static obstacle boundary or the predicted trajectory of the dynamic obstacle, the candidate passable area is corrected with the feasible time window of the path segment and the UAV dynamics conditions as constraints, with the goal of maximizing the passable space of the corridor, to obtain a corrected local safe corridor; the local safe corridor is spliced ​​together according to the path segment sequence to form a globally continuous safe corridor channel.

[0015] In a preferred embodiment, the step of outputting the trajectory with the safety corridor as a constraint is as follows: A multi-objective trajectory optimization problem is constructed by using the local safety corridor within each path segment as a constraint; the multi-objective trajectory optimization problem is transformed into a piecewise quadratic programming problem to obtain optimized local trajectory segments; the optimized local trajectory segments are spliced ​​together in the path sequence to form a complete continuous trajectory; a time consistency check is performed on the spliced ​​trajectory, and if conflicts exist, the local velocity profile is adjusted first; obstacle perception data during the UAV's flight is acquired in real time, and the safety corridor boundaries of the corresponding path segments are updated to form new local safety corridors; for the updated safety corridor area, local trajectory optimization is re-executed, and online trajectory correction is performed using a short-time MPC algorithm.

[0016] A path planning system for a quadcopter unmanned aerial vehicle (UAV) includes an initial path module, a correction module, and an output module, which are interconnected. The initial path module is used to acquire perception data of the target area, build a multi-resolution map, and generate an initial path based on the multi-resolution map. The correction module is used to correct the initial path and construct a safe corridor. The output module is used to output the trajectory with the safe corridor as a constraint.

[0017] The technical effects and advantages of the path planning method and system for a quadcopter unmanned aerial vehicle (UAV) of the present invention are as follows:

[0018] 1. This invention systematically corrects and optimizes the initial path. By traversing the node sequence, redundant nodes are deleted and time windows are inherited. Simultaneously, at key points with large turning angles, continuous curvature curves are used instead of broken lines to smooth the path. Based on this, dynamic constraints are introduced to construct feasibility boundaries for path segments, transforming the path optimization problem into a convex optimization problem. A quadratic programming method is used to optimize the control points, yielding locally corrected curves. Furthermore, through bidirectional time window checks and velocity profile adjustments, safe correction and feasibility maintenance of the path under the influence of dynamic obstacles are achieved, thereby significantly improving the reliability and flight efficiency of the path.

[0019] 2. This invention proposes a method for constructing safe corridors. Based on a modified initial path, the path is divided into segments according to key nodes. Under the obstacle distribution of the local map, multi-scale expansion is used to generate candidate passage areas. These candidate areas are then modified by combining UAV dynamics conditions and time window constraints to form local safe corridors. These local safe corridors are then stitched together to form a globally continuous safe passage, providing clear constraint boundaries for trajectory optimization. This enables the UAV to stably avoid obstacles during flight and maximizes the passage space. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the path planning method for a quadcopter drone according to the present invention.

[0021] Figure 2 This is a schematic diagram of the path planning system structure of a quadcopter unmanned aerial vehicle according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1, Figure 1 The present invention provides a path planning method for a quadcopter unmanned aerial vehicle, comprising the following steps:

[0024] S1, acquire perception data of the target area, build a multi-resolution map, and generate an initial path based on the multi-resolution map;

[0025] In this embodiment, perception data of the target area is acquired, a multi-resolution map is built, and an initial path is generated based on the multi-resolution map, as detailed below:

[0026] Acquire perception data of the target area, including point cloud and image data and location information collected by lidar and depth camera;

[0027] Based on the perception data, a multi-resolution map combining global and local data is established. The global map uses a low-resolution grid or topological skeleton to represent the overall structure of the target area and the distribution of large obstacles. The local map uses a high-resolution voxel grid to represent the fine obstacle information around the UAV. Confidence labels are added to the obstacles in the map to characterize perception uncertainty. For dynamic obstacles, a time-series prediction model is used to generate short-term motion trajectories, and the prediction results are updated in the map.

[0028] The global map and local maps are linked through an index relationship, allowing switching between different resolution levels during the route planning process;

[0029] Based on mission requirements, determine the starting and target positions of the UAV on the global map, and identify key access directions by combining the global topology to narrow down the subsequent search range and avoid ineffective exploration.

[0030] Based on the global and local maps, and combined with the starting and target positions of the UAV, the search space of the UAV is constrained, and initial path candidate solutions are generated within the constrained search space.

[0031] By using the principle of curvature continuity, the initial path candidate solution is spliced ​​between segments to form a geometric path that meets the continuity requirements. This ensures that the path is continuous in position, direction, and curvature, avoiding sharp turns or situations that do not meet the dynamic constraints.

[0032] Based on the geometric path, an estimated arrival time window is allocated at its key nodes. The key nodes are turning points or points with significant speed changes formed during the path splicing process. The time window is determined by a combination of the UAV's speed boundary conditions, the prediction results of dynamic obstacles, and the local environmental complexity.

[0033] By imposing time parameterization constraints on key nodes, the geometric path is transformed into an initial path with temporal attributes.

[0034] In this embodiment, based on the global map and local map, and combined with the UAV's starting and target positions, the UAV's search space is constrained, and initial path candidate solutions are generated within the constrained search space, as follows:

[0035] In the global map, the mission area of ​​the UAV is determined with the starting position and the target position as the boundary. Combined with the obstacle distribution information in the local map, the mission area is cropped to remove obviously impassable spaces and generate a constrained search space.

[0036] The search space is divided into regions, forming areas with dense obstacles and open areas;

[0037] When there are densely packed areas of obstacles in the search space, based on the spatial distribution information of the obstacles, the sampling probability of nodes in narrow passages or high-risk areas is increased through an importance sampling mechanism to obtain a set of sampling points in dense areas.

[0038] When there are open areas in the search space, a uniform sampling strategy is adopted to distribute the sampling nodes evenly in the open areas, avoid missing potential low-cost paths, and ensure the coverage of the overall search process, thus obtaining a set of sampling points in the open areas.

[0039] The dense region sampling point set and the open region sampling point set are merged to construct a unified candidate node set. Then, using the candidate node set as input, several coarse path candidate solutions are generated through a path search algorithm.

[0040] In this embodiment, the steps for obtaining the expected arrival time window are as follows:

[0041] Based on the maximum and minimum speeds of the UAV and the geometric length of the path segments, the theoretical shortest and longest flight times for each path segment are calculated to obtain the initial time interval for that path segment.

[0042] By using a kinematic prediction model, the future trajectory of dynamic obstacles is estimated. Near key nodes, if it is predicted that an obstacle may occupy a relevant path segment, the time period that conflicts with the obstacle is eliminated in the initial time interval, thereby narrowing the feasible time range and forming the time interval constrained by the obstacle.

[0043] The time interval constrained by obstacles is corrected based on the local environmental complexity, which includes obstacle density and flight space narrowness.

[0044] When the environment is highly complex, the time interval should be appropriately extended and redundancy increased to reduce control risks.

[0045] When the environment complexity is low, maintain a tighter time interval to improve path execution efficiency;

[0046] The revised time interval is used as the expected arrival time window for the critical node, and the time windows of adjacent critical nodes are coordinated to ensure the continuity of the overall path in the time dimension.

[0047] S2, correct the initial path and construct a safe corridor;

[0048] In this embodiment, the initial path is corrected as follows:

[0049] Traverse the node sequence of the initial path. When the directional change between two adjacent path segments is less than the preset geometric threshold and there is no conflict in the corresponding time window, delete the redundant intermediate nodes. At the same time, merge and inherit the time window of the deleted node with the nodes before and after it to ensure that the path simplification operation does not destroy the global temporal consistency.

[0050] Key points in the path whose turning angle exceeds a threshold are detected, and the original polyline is replaced with a smooth curve segment with continuous curvature (including cubic spline, Clothoid curve or Bézier curve) to obtain the initial local curve segment.

[0051] Based on the geometric length of the replaced curve segment, recalculate the shortest and longest flight times for that segment and compare them with the corresponding time windows;

[0052] When the recalculation time is within the original time window, update the node's time window;

[0053] If the recalculation result exceeds the time window, the following measures will be executed in order of priority:

[0054] a. Relax the time window within permissible limits;

[0055] b. Insert a waiting action in the adjacent path segment;

[0056] c. If it is still not feasible, mark the path segment as uncorrectable and trigger candidate path regeneration;

[0057] Acquire the dynamic data of the UAV and use the dynamic parameters as constraints to establish the feasibility boundary of the path segment. The dynamic data includes acceleration, angular velocity and rate of climb, and the feasibility boundary of the path segment includes maximum acceleration, maximum angular velocity and maximum rate of climb.

[0058] Using the initial local curve segment as the initial solution, and combining the feasibility boundary of the path segment, a multi-objective optimization function is constructed. The multi-objective optimization function includes minimizing the rate of curvature change, minimizing the control input energy, and minimizing the offset from the original path.

[0059] The multi-objective optimization function is transformed into a convex optimization problem, and the control points of the path segment are optimized and solved using a quadratic programming method to obtain an optimized local correction curve. The local correction curve has the following characteristics: the control points are optimized and adjusted so that the correction curve remains geometrically continuous with the adjacent path segments; the curve parameterization function satisfies the acceleration and angular velocity boundary conditions of the UAV; and velocity profiles and acceleration profiles matching the corresponding time windows are generated, thereby ensuring the feasibility of the path in both the time and dynamic dimensions.

[0060] After obtaining the local correction curve, a two-way check is performed on the time window of the entire path from top to bottom and from bottom to top to ensure that the arrival time of the global path remains monotonically non-decreasing and does not conflict with the predicted occupancy interval of dynamic obstacles.

[0061] When a conflict is detected, the velocity profile is adjusted first, and the local velocity is increased or decreased when the boundary conditions allow.

[0062] If the problem cannot be resolved, insert a wait or fall back to the candidate path set.

[0063] The local correction curves are spliced ​​together in the path order to obtain the corrected initial path;

[0064] The corrected initial path is projected into the time-space domain, and collision detection of static and dynamic obstacles is performed. If the verification passes, the output is used as the final path.

[0065] In this embodiment, a safe corridor is constructed as follows:

[0066] The initial path is divided into multiple path segments according to key nodes. Each path segment corresponds to a clear start and end point and an expected arrival time window, which serves as the basic unit for subsequent corridor construction.

[0067] For each path segment, combined with the distribution of obstacles in the local map, a multi-scale dilation method is used to generate candidate passage areas around the path segment. The area adjacent to the path segment is dilated with high resolution to ensure boundary accuracy, while the area far from the path is dilated with low resolution to reduce computational complexity.

[0068] If the candidate area conflicts with the static obstacle boundary or the predicted trajectory of the dynamic obstacle, the candidate passage area is corrected by taking the feasible time window of the path segment and the UAV dynamics conditions as constraints and the goal of maximizing the passable space of the corridor, so as to obtain the corrected local safe corridor.

[0069] Local safety corridors are spliced ​​together according to the path segment sequence. During the splicing process, it is ensured that adjacent corridors overlap in the spatial dimension and have intersecting time windows in the temporal dimension, thereby forming a globally continuous safety corridor channel.

[0070] S3, with the safety corridor as the constraint, outputs the trajectory;

[0071] In this embodiment, the trajectory is output using the safety corridor as a constraint, as follows:

[0072] Using the local safety corridor within each path segment as a constraint, a multi-objective trajectory optimization problem is constructed. The multi-objective trajectory optimization problem includes minimizing the rate of change of curvature, minimizing instantaneous acceleration and angular velocity, and minimizing deviation from the initial path.

[0073] The multi-objective trajectory optimization problem is transformed into a piecewise quadratic programming problem, resulting in optimized local trajectory segments.

[0074] The optimized local trajectory segments are spliced ​​together in the path sequence to form a complete continuous trajectory.

[0075] Perform a time consistency check on the spliced ​​trajectory to ensure that the global path arrival time monotonically increases and does not conflict with the time occupied by dynamic obstacles.

[0076] If a conflict exists, the local velocity profile should be adjusted first.

[0077] During the flight of the UAV, obstacle perception data is acquired in real time, the safe corridor boundary of the corresponding path segment is updated, and a new local safe corridor is formed. The safe corridor boundary is defined as follows: if the candidate area conflicts with the static obstacle boundary or the predicted trajectory of the dynamic obstacle, the feasible time window of the path segment and the UAV dynamic conditions are used as constraints, with the goal of maximizing the passable space of the corridor.

[0078] For the updated safe corridor area, local trajectory optimization is re-executed to ensure that the trajectory always lies within the legal passage space, and online trajectory correction is performed through a short-time MPC algorithm to achieve dynamic obstacle avoidance.

[0079] Example 2, Figure 2 The present invention provides a system for path planning of a quadcopter UAV, comprising an initial path module, a correction module, and an output module, wherein the modules are interconnected.

[0080] The initial path module is used to acquire perception data of the target area, build a multi-resolution map, and generate an initial path based on the multi-resolution map.

[0081] The correction module is used to correct the initial path and build a safe corridor;

[0082] The output module is used to output the trajectory with the safety corridor as a constraint.

[0083] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0084] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0085] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0086] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0088] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A path planning method for a quadcopter unmanned aerial vehicle (UAV), characterized in that, include: Acquire sensor data of the target area, build a multi-resolution map, and generate an initial path based on the multi-resolution map; The initial path is corrected, and a safe corridor is constructed; Output the trajectory, using the safety corridor as a constraint. The process of acquiring perception data of the target area, building a multi-resolution map, and generating an initial path based on the multi-resolution map is as follows: Acquire perception data of the target area, and build a multi-resolution map based on the perception data. The multi-resolution map includes a global map and a local map. The global map uses a low-resolution grid or topological skeleton to represent the overall structure of the target area and the distribution of large obstacles. The local map uses a high-resolution voxel grid to represent the fine obstacle information around the UAV. Add confidence labels to obstacles in the map to characterize perception uncertainty. Use a time series prediction model to generate short-term motion trajectories for dynamic obstacles. Based on mission requirements, determine the starting and target positions of the UAV in the global map, and identify key access directions by combining the global topology. Based on the global and local maps, and combined with the starting and target positions of the UAV, the search space of the UAV is constrained, and initial path candidate solutions are generated within the constrained search space. By using the principle of curvature continuity, the initial path candidate solutions are spliced ​​together segment by segment to form a geometric path that meets the continuity requirement; Based on the geometric path, allocate the expected arrival time window at key nodes; By imposing time parameterization constraints on key nodes, the geometric path is transformed into an initial path with temporal attributes. The steps for obtaining the expected arrival time window are as follows: Based on the maximum and minimum speeds of the UAV and the geometric length of the path segments, the theoretical shortest and longest flight times for each path segment are calculated to obtain the initial time interval for that path segment. By using a kinematic prediction model, the future trajectory of dynamic obstacles is estimated. Near key nodes, if it is predicted that an obstacle occupies a relevant path segment, the time period that conflicts with the obstacle is eliminated in the initial time interval, thereby narrowing the feasible time range and forming the time interval constrained by the obstacle. The time interval after obstacle constraints is adjusted according to the complexity of the local environment. The revised time interval will be used as the expected arrival time window for this critical node; The construction of the secure corridor is detailed as follows: The initial path is divided into multiple path segments based on key nodes, and each path segment corresponds to a clear start and end point and an estimated arrival time window. For each path segment, candidate passage areas are generated around the path segment by combining the distribution of obstacles in the local map and using a multi-scale dilation method. If the candidate area conflicts with the static obstacle boundary or the predicted trajectory of the dynamic obstacle, the candidate passage area is corrected by taking the feasible time window of the path segment and the UAV dynamics conditions as constraints and the goal of maximizing the passable space of the corridor, so as to obtain the corrected local safe corridor. The local safety corridors are spliced ​​together according to the path segment sequence to form a globally continuous safety corridor channel; The trajectory is output under the constraint of a safe corridor, as detailed below: By using the local safety corridors within each path segment as constraints, a multi-objective trajectory optimization problem is constructed. The multi-objective trajectory optimization problem is transformed into a piecewise quadratic programming problem, resulting in optimized local trajectory segments. The optimized local trajectory segments are spliced ​​together in the path sequence to form a complete continuous trajectory. Perform a time consistency check on the spliced ​​trajectory. If a conflict exists, prioritize adjusting the local velocity profile. Real-time acquisition of obstacle perception data during drone flight, updating the safety corridor boundary of the corresponding path segment, and forming a new local safety corridor; For the updated safe corridor area, local trajectory optimization is re-executed, and online trajectory correction is performed using a short-time MPC algorithm.

2. The path planning method for a quadcopter UAV according to claim 1, characterized in that, The search space of the UAV is constrained based on the global and local maps, combined with the UAV's starting and target positions, and initial path candidate solutions are generated within the constrained search space, as detailed below: In the global map, the mission area of ​​the UAV is determined with the starting position and the target position as the boundary. Combined with the obstacle distribution information in the local map, the mission area is clipped to generate a constrained search space. The search space is divided into regions, forming areas with dense obstacles and open areas; When there are densely packed areas of obstacles in the search space, based on the spatial distribution information of the obstacles, the sampling probability of nodes in the narrow channel area is increased through the importance sampling mechanism to obtain the sampling point set of the dense area. When there is an open area in the search space, a uniform sampling strategy is adopted to make the sampling nodes evenly distributed in the open area, thus obtaining the open area sampling point set. The dense region sampling point set and the open region sampling point set are merged to construct a unified candidate node set. Then, using the candidate node set as input, several coarse path candidate solutions are generated through a path search algorithm.

3. The path planning method for a quadcopter UAV according to claim 1, characterized in that, The initial path is corrected as follows: Traverse the node sequence of the initial path. When the directional change between two adjacent path segments is less than the preset geometric threshold and there is no conflict in the corresponding time window, delete the redundant intermediate node. At the same time, merge and inherit the time window of the deleted node with the nodes before and after it. Key points in the path whose turning angle exceeds a threshold are detected, and the original polyline is replaced with a smooth curve segment with continuous curvature to obtain an initial local curve segment; Based on the geometric length of the replaced curve segment, recalculate the shortest and longest flight times of the segment, compare them with the corresponding time windows, and update the time windows of the nodes. Acquire the dynamic data of the UAV and use the dynamic parameters as constraints to establish the feasibility boundary of the path segment; Using the initial local curve segment as the initial solution, and combining the feasibility boundary of the path segment, a multi-objective optimization function is constructed; The multi-objective optimization function is transformed into a convex optimization problem, and the curve control points of the path segment are optimized and solved using the quadratic programming method to obtain the optimized local correction curve. After obtaining the local correction curve, a bidirectional check is performed on the time window of the entire path; When a conflict is detected, the velocity profile is adjusted first, and the local velocity is increased or decreased when the boundary conditions allow. If the problem cannot be resolved, insert a wait or fall back to the candidate path set. The local correction curves are spliced ​​together in the path order to obtain the corrected initial path; The corrected initial path is projected into the time-space domain, and collision detection of static and dynamic obstacles is performed. If the verification passes, the output is used as the final path.

4. A system using the path planning method for a quadcopter unmanned aerial vehicle as described in any one of claims 1-3, characterized in that, It includes an initial path module, a correction module, and an output module, and the modules are connected to each other; The initial path module is used to acquire perception data of the target area, build a multi-resolution map, and generate an initial path based on the multi-resolution map. The correction module is used to correct the initial path and build a safe corridor; The output module is used to output the trajectory with the safety corridor as a constraint.

Images