Unmanned aerial vehicle autonomous cruise and obstacle avoidance method and system for dynamic no-fly zone

Abstract

The invention discloses an unmanned aerial vehicle autonomous cruise and obstacle avoidance method and system for a dynamic no-fly zone. Relates to the technical field of unmanned aerial vehicle autonomous navigation and path planning. The method comprises the following steps: issuing and keeping a no-fly zone set static in a task initialization stage; fusing the multi-source sensing data through a layered SLAM mapping module to generate an occupied grid map; performing global path planning by adopting an improved BFS algorithm and a weighted greedy strategy based on a map, and generating a patrol sequence covering reachable points; compressing and smoothing the spliced path to generate a final cruise trajectory; and triggering local trajectory re-planning for sudden obstacles in a non-no-fly zone based on SLAM perception in flight. According to the invention, autonomous cruise and dynamic obstacle avoidance with high coverage rate and low redundancy are realized under the condition of strictly following the constraint of the static no-fly zone.

Description

Technical Field

[0001] This invention relates to the field of autonomous navigation and path planning technology for unmanned aerial vehicles (UAVs), specifically to a method and system for autonomous cruising and obstacle avoidance of UAVs in dynamic no-fly zones. Background Technology

[0002] With the increasing prevalence of drones in various fields, achieving safe and autonomous navigation in complex environments has become a critical issue. Traditional GPS-based navigation technology struggles to provide reliable support in weak signal or dynamic environments, while visual SLAM technology, although capable of GPS-free positioning, also faces limitations, particularly in meeting spatial constraints such as no-fly zones, where technical challenges arise.

[0003] Specifically, current technology faces the following problems: Navigation issues in weak signal or dynamic environments: Traditional GPS-based navigation methods cannot ensure stable positioning of drones when signals are weak or the environment is dynamically changing (such as tall buildings, tunnels, dense buildings, etc.). Therefore, it is necessary to rely on other positioning technologies to replace or supplement GPS navigation.

[0004] While visual SLAM technology can provide GPS-free positioning, its limitation lies in its inability to simultaneously meet spatial constraints such as no-fly zones. No-fly zones are a critical constraint in flight missions, but visual SLAM focuses more on mapping and localization, making it difficult to effectively consider and avoid the existence of no-fly zones when planning flight paths.

[0005] No-fly zones are typically established by the management authority before a mission and remain unchanged during flight. Failure to adequately consider these no-fly zone constraints during mission planning may lead to conflicts between the flight mission and the no-fly zone, or even cause flight accidents. Current technology lacks an effective method to fix no-fly zone constraints during mission initialization and dynamically adhere to them during flight.

[0006] Efficiently combining no-fly zone constraints and dynamic flight environments (such as temporary obstacles and weather changes) to generate the optimal cruise path during flight mission planning is a challenge. Existing path planning methods often suffer from problems such as path redundancy, insufficient coverage, and poor obstacle avoidance performance. Therefore, innovative path planning algorithms are needed to ensure flight efficiency, obstacle avoidance performance, and compliance with no-fly zone constraints.

[0007] In actual flight, sudden obstacles or unforeseen flight environments may affect the execution of the planned path. Existing technologies often react slowly to sudden obstacles and cannot quickly replan the path, leading to reduced flight safety. Therefore, a technology capable of rapid local path replanning is needed to cope with sudden obstacles in flight and ensure the continuity of flight missions.

[0008] In summary, existing technologies have certain shortcomings in addressing no-fly zone constraints, path planning and obstacle avoidance, and adaptability to dynamic environments. Therefore, a new approach is needed to solve these technical problems. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides a method and system for autonomous navigation and obstacle avoidance of unmanned aerial vehicles (UAVs) in dynamic no-fly zones, thereby solving the technical problems mentioned in the background section.

[0010] To achieve the above objectives, the present invention provides the following technical solution: An autonomous navigation and obstacle avoidance method for unmanned aerial vehicles (UAVs) in dynamic no-fly zones includes the following steps: S1. Mission Initialization and Static No-Fly Zone Setting: During the mission initialization phase, a set of no-fly zones is generated by the ground control terminal or mission scheduling center and sent to the UAV system; the set of no-fly zones remains static throughout the entire duration of this UAV flight mission and does not receive any modifications or updates. S2. Environmental perception and hierarchical SLAM mapping: Environmental data is collected by the UAV’s onboard visual sensors and inertial measurement unit and input into the hierarchical SLAM mapping module for tight coupling processing. The UAV’s pose is estimated in real time and the occupied grid map is constructed and updated. The grid corresponding to the no-fly zone set is permanently marked as impassable in the occupied grid map. S3. Grid modeling and reachable point set generation: Based on the occupied grid map, the patrol area is discretized into a regular grid, grids occupied by obstacles and grids permanently marked as impassable are removed, and all passable grid points are extracted to form a reachable point set. S4. Global Patrol Path Planning: Based on the set of passable grid points, global path planning is performed, the process of which includes: S41, BFS Local Shortest Path Search: An improved breadth-first search algorithm is used to calculate the shortest feasible path for any two passable grid points in the reachable point set; S42. Global inspection sequence generation: Starting from the task start point, a weighted greedy strategy is used to select the next inspection point in sequence to generate a global access sequence covering the set of reachable points. S43. Path splicing: The local shortest path segments between adjacent points in the global access sequence are sequentially spliced ​​to form the initial cruise trajectory; S44. Trajectory Generation: The initial cruise trajectory is compressed and smoothed, including deleting continuous intermediate nodes with the same direction, sampling straight line segments with lengths exceeding a threshold at equal intervals to retain key waypoints, and using a smooth interpolation algorithm to generate the final cruise trajectory that meets the speed and acceleration continuity requirements of the flight control system. S5. Trajectory Distribution and Autonomous Cruise: The final cruise trajectory is distributed to the flight control system to control the UAV to perform autonomous cruise mission; S6. Dynamic obstacle avoidance and local trajectory replanning: During autonomous cruise, if the hierarchical SLAM mapping module detects a sudden obstacle in a non-no-fly zone and affects the current trajectory, local trajectory replanning is triggered. Based on the current UAV position and obstacle information, a new waypoint is selected in a local area and a new local cruise trajectory is generated. The new local cruise trajectory is seamlessly integrated into the currently executing global cruise trajectory, while ensuring that the no-fly zone set remains unchanged.

[0011] As a further aspect of the present invention, the process of generating a new local cruise trajectory includes: S61. Local path calculation: Based on the newly selected waypoint, the improved breadth-first search algorithm is used to calculate the local shortest feasible path from the current UAV position to the new waypoint, and between the new waypoints. S62, Local Path Concatenation: The calculated local shortest feasible paths are sequentially concatenated to form a local initial path; S63. Local trajectory optimization: The local initial path is compressed and smoothed to generate the new local cruise trajectory.

[0012] As a further aspect of the present invention, the execution process of the hierarchical SLAM mapping module includes: Visual features are extracted from image frames through a feature extraction layer and time-aligned with IMU data. A local occupying raster map is constructed based on the aligned data by creating local layers, and local constraints are submitted to the global optimization layer. A global optimization layer is used to perform global correction on UAV pose and map based on loop closure detection and tight coupling optimization to reduce cumulative errors.

[0013] As a further aspect of the present invention, the improved breadth-first search algorithm specifically includes: Before searching for the path between the starting point and the target point, a connectivity analysis is performed on the set of reachable points, dividing it into several connected subgraphs. Determine whether the starting point and the target point belong to the same connected subgraph; If they do not belong to the same subgraph, the path reachability is evaluated based on the pre-computed boundary connectivity cost table or a cross-subgraph merge search is triggered. If they belong to the same subgraph, the standard BFS algorithm is executed in the four neighborhood directions to output the shortest path sequence.

[0014] As a further aspect of the present invention, in the weighted greedy strategy, the comprehensive weight function w(p) used to select the next inspection point is defined as: ; Among them, S k Let p be the current location and p be the candidate target point, dist(S k ,p) represents the Manhattan distance between the current point and the candidate target point, and q (p) a is the importance coefficient of candidate target points preset based on task requirements. (p) The accessibility score is calculated based on the historical access status and access frequency of the path. α, β, and γ are preset adjustable non-negative weight coefficients, and satisfy α+β+γ=1.

[0015] As a further aspect of the present invention, the smooth interpolation algorithm employs spline curves or fifth-order polynomials to generate a smooth trajectory that meets the speed and acceleration continuity requirements of the UAV flight control system.

[0016] An autonomous navigation and obstacle avoidance system for unmanned aerial vehicles (UAVs) in dynamic no-fly zones includes: The ground control terminal is used to issue the set of no-fly zones during the mission initialization phase and to receive the status information of the UAV during the mission. The perception module is used to collect environmental data through a visual sensor and an inertial measurement unit; The hierarchical SLAM mapping module is used to run the hierarchical SLAM mapping steps, generate and maintain the occupied grid map; The path planning module is used to execute the global patrol path planning steps and generate the initial patrol trajectory; The trajectory optimization module is used to perform the trajectory generation step and generate the final cruise trajectory; The flight control execution module is used to receive and execute the final cruise trajectory and control the UAV to complete the cruise mission; The communication module is used to transmit data between the UAV and the ground control terminal; The ground control terminal is used to issue the set of no-fly zones during the mission initialization phase and to receive the status information of the UAV during the mission.

[0017] As a further aspect of the present invention, the communication module includes a redundant communication unit, which supports dual-path redundant switching between low-latency short-range wireless links and public cellular links to ensure reliable transmission of task data and updated maps in complex environments.

[0018] As a further embodiment of the present invention, the perception module further integrates an edge behavior recognition unit. This unit performs real-time analysis of the visual data stream based on airborne computing resources to identify specific target behaviors, and uses the recognition results as a dynamic input of the target importance coefficient, feeding them back to the path planning module to adjust the patrol priority in real time.

[0019] As a further aspect of the present invention, the path planning module has a built-in event triggering mechanism, which is used to generate a new local cruise trajectory by performing local path calculation, local path splicing and local trajectory optimization when the dynamic obstacle avoidance and local trajectory replanning are triggered, and to send it to the flight control execution module first for seamless trajectory access.

[0020] Compared with the prior art, the beneficial technical effects of the present invention are as follows: 1. By setting the no-fly zone set as a global static constraint in the mission initialization phase and prohibiting its modification throughout the flight, all path planning is carried out under strict and clear spatial constraints, fundamentally eliminating the risk of intruding into the no-fly zone due to accidental modification of the no-fly zone during flight or neglect of constraints during planning; 2. The hierarchical SLAM mapping technology that integrates vision and inertial sensing can achieve stable and high-precision real-time positioning and map building in environments without GPS or with weak signals. This frees the UAV from dependence on fixed navigation signals and enables it to have true fully autonomous cruising capabilities in complex and unknown environments. 3. By combining an improved breadth-first search algorithm with a weighted greedy strategy, a global patrol sequence is intelligently generated while ensuring the shortest path. This is achieved by comprehensively considering distance, task priority, and point reachability. The combination of these two approaches significantly reduces path redundancy and total patrol cost. 4. By using path compression and spline interpolation, a smooth trajectory that conforms to flight control dynamics constraints is generated, reducing unnecessary waypoints and sharp turns, and improving flight continuity and energy efficiency; 5. In the event of sudden obstacles in non-no-fly zones, the system can quickly trigger local trajectory replanning and achieve seamless trajectory replacement without interrupting the mission or modifying the global configuration, thus ensuring the continuity and stability of flight. Attached Figure Description

[0021] Figure 1 This is a general flowchart of the method described in the embodiments of the present invention; Figure 2This is a block diagram of the system described in an embodiment of the present invention; Figure 3 This is a flowchart of the global patrol path planning in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the dynamic obstacle avoidance and local trajectory replanning of an embodiment of 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1 like Figure 1 As shown, this invention provides a method for autonomous navigation and obstacle avoidance of unmanned aerial vehicles (UAVs) in dynamic no-fly zones, including the following steps: S1. Task Initialization and Static No-Fly Zone Setting During the mission initialization phase, the ground control terminal or mission scheduling center generates and issues a no-fly zone set B to the UAV system. The no-fly zone set B remains static throughout the entire duration of this UAV flight mission and does not receive any modifications or updates. The no-fly zone set B contains several no-fly zones, each represented by its geographical coordinate range or grid area.

[0024] S2, Environmental Perception and Hierarchical SLAM Mapping After the drone takes off, it synchronously acquires environmental images and attitude information through the onboard visual sensors and inertial measurement units (IMUs) and inputs them into the hierarchical SLAM mapping module for tight coupling processing. The drone's pose is estimated in real time and the occupied grid map is constructed and updated. The grid corresponding to the no-fly zone set is permanently marked as impassable in the occupied grid map. The execution process of the hierarchical SLAM mapping module includes: Visual features are extracted from image frames through a feature extraction layer and time-aligned with IMU data. A local occupying raster map is constructed based on the aligned data by creating local layers, and local constraints are submitted to the global optimization layer. A global optimization layer is used to perform global correction on UAV pose and map based on loop closure detection and tight coupling optimization to reduce cumulative errors.

[0025] S3, Mesh Modeling and Reachable Point Set Generation Based on the occupied grid map, the task patrol area is discretized into an R×C regular grid; Let the set of no-fly zones issued by the ground control during the mission initialization phase be B = {b1, b2, ..., b}. m}, where each no-fly zone b j It is represented by its grid point set or polygon coordinates in the raster coordinate system; the occupancy function (initial no-fly zone mapping) is defined as: ; Where grid(r,c)=0 indicates that the grid point belongs to the no-fly zone and is permanently marked as impassable; grid(r,c)=1 indicates that the grid point is initially passable; during the mission B, it remains static and does not receive new no-fly zone assignments during flight. Subsequently, the system traverses the entire grid and, based on the latest occupied grid map sensed and maintained in real time by the hierarchical SLAM module, filters out all currently passable grid points to form a reachable point set G; specifically, the reachable point set G consists of grid points that satisfy the following conditions: G={(r,c)|1≤r≤R,1≤c≤C, and grid(r,c)=1}; Note: The reachable point set G here is generated only based on the static no-fly zone and is used for the subsequent global path planning step S4. Dynamic obstacles that may appear during flight are perceived in real time by the hierarchical SLAM module and processed in the dynamic obstacle avoidance and local trajectory replanning step S6.

[0026] S4, Global Patrol Path Planning like Figure 3 As shown, based on the set of passable grid points, global path planning is performed, and the process includes: S41, BFS (Local Shortest Path Search) An improved breadth-first search (BFS) algorithm is used to calculate the shortest feasible path for any two walkable grid points in the reachable point set G; The core of the improved breadth-first search algorithm lies in the introduction of connectivity preprocessing and intelligent decision-making mechanisms, specifically including: Connectivity analysis: Before starting the path search, a connectivity analysis is performed on the set of reachable points G, dividing it into several connected subgraphs {C1, C2, ...}. Reachability prediction: Given a starting point S=(r s ,c s ) and target point T=(r t ,c t First, determine whether the two belong to the same connected subgraph; Intelligent search decision-making: If the starting point S and the target point T do not belong to the same subgraph, the overall reachability of the path is evaluated by querying the pre-computed boundary connectivity cost table. The decision is made on whether to trigger a cross-subgraph merge search or directly return "unreachable" information, thereby avoiding an invalid full graph search. If the starting point S and the target point T belong to the same subgraph, then the standard BFS algorithm is executed in the four-neighbor or eight-neighbor direction to find the shortest path. Taking four-neighbor search as an example, the process is as follows: Define the set of four neighborhood directions: D = {(-1,0), (1,0), (0,-1), (0,1)}; Given a starting point S=(r s ,c s ) and target point T=(r t ,c t Given S, T∈G; and satisfying 1≤r≤R, 1≤c≤C, and grid(r,c)=1, the BFS algorithm is used to expand on the four-neighborhood to find the shortest feasible grid path from S to T. For reachable cases, the local shortest path length is defined as: |L(S,T)|=min{k|There exists a grid path of length k from S to T}; The output of BFS is a path sequence: L(S,T)=((r (0) ,c (0) ),(r (1) ,c (1) ),...,(r (L) ,c (L) )); Where, (r (0) ,c (0) )=S,(r (L) ,c (L) )=T; S42, Global Patrol Sequence Generation Starting from the task origin S0, a weighted greedy strategy is used to sequentially select the next inspection point, generating a global access sequence {S0, S1, ..., S...} covering the set of reachable points G. m}; Let the set of all walkable lattice points in the reachable point set G be denoted as {P1, P2, ..., P...} N}, where N refers to the total number of traversable grid points in the reachable point set G, and P i =(r i ,c i ), r i and c iThese represent the row and column numbers of the point in the raster map, respectively. To form a comprehensive inspection sequence, the Manhattan distance between points is introduced: dist(P i ,P j )=|r i -r j |+|c i -c j |; To consider task priority and reachability, a comprehensive weighting function is used in the greedy selection process. Let the current point be S. k (S) k ∈{P1,P2,...,P N Candidate target point is p (p∈{P1,P2,...,P)). N Then the weight function w(p) is defined as: ; Where, q (p) >0 represents a preset target importance coefficient (e.g., based on ecological protection weights or manual labeling); a (p) >0 represents the reachability score (reflecting path accessibility, historical access frequency, etc.); α, β, γ are adjustable non-negative weight coefficients that satisfy α+β+γ = 1 and can be adjusted according to task scheduling; The greedy selection rule is: from all unvisited passable points, select the point with the smallest overall weight w(p) as the next patrol point, that is: ; Among them, S k+1 Indicates the next checkpoint to be visited; p∈G\{S0,S1,...,S k} indicates that the range of values ​​for p is all unvisited passable points; This process continues until the predetermined target set is covered or the stopping condition is met, for each pair of adjacent access points (S) i ,S i+1 The algorithm for finding the shortest path in step S41 (improved BFS) is invoked to calculate the shortest feasible path sequence L(S). i ,S i+1 ); S43, Path splicing The local shortest path segments between adjacent points in the global access sequence are sequentially concatenated to form the initial cruise trajectory P: ; S44, Trajectory Generation To reduce control and communication overhead during flight, path compression rules are introduced: Compression: If three consecutive points on the path satisfy the condition of consistent direction (rt , c t ), (r t+1 , c t+1 ), (r t+2 , c t+2 If the direction consistency condition is satisfied (i.e., the vector directions are the same), then the intermediate node (r) will be... t+1 , c t+1 )delete; Sampling and smoothing: For any continuous straight line segment, if its length l satisfies l>L th If a preset threshold is set, equally spaced sampling is performed to retain key waypoints, and spline curves or fifth-order polynomial interpolation are used to generate smooth trajectories to meet the requirements for flight control speed / acceleration continuity. The final compressed and smoothed cruise trajectory P is obtained. * : .

[0027] S5, trajectory distribution and autonomous navigation The final cruise trajectory P * The command is sent to the flight control system to control the drone to perform autonomous cruise missions. The flight control system executes the cruise mission step by step according to the planned waypoints, and automatically maintains a safe distance at the boundary of the no-fly zone to ensure that the drone does not enter the preset no-fly zone during flight.

[0028] S6, Dynamic Obstacle Avoidance and Local Trajectory Replanning During autonomous cruise, the hierarchical SLAM mapping module continuously performs local environmental perception and updates the occupied grid (this update does not include any changes to the no-fly zone set B), thereby always ensuring flight compliance; If the hierarchical SLAM mapping module detects a sudden obstacle in a non-no-fly zone that affects the current execution trajectory (such as a temporarily parked vehicle, moving personnel, or an outstretched tree branch), it will trigger a local trajectory replanning. like Figure 4 As shown, the process of generating a new local cruise trajectory includes: S61, Local Path Calculation Using the UAV's current location as the new planning starting point, and based on the real-time obstacle location and size information provided by the hierarchical SLAM module, the next (or a group of) temporary waypoints are re-evaluated and selected within the local neighborhood of the affected area. Subsequently, an improved breadth-first search (BFS) algorithm is used to calculate the local shortest feasible paths from the current UAV location to the new waypoints, as well as between the new waypoints. This process also strictly avoids all no-fly zones and newly discovered obstacles. S62, Local Path Concatenation The one or more local shortest feasible paths calculated in step S61 are spliced ​​together in the order of being visited to form a complete but not yet optimized local initial path; this path can geometrically connect the starting point and the target point, but may contain redundant turns. S63, Local Trajectory Optimization For this local initial path, the same compression and smoothing process as step S44 in the global planning is performed. First, the same "direction consistency" rule is applied to delete collinear intermediate nodes. Second, long straight segments are sampled at equal intervals to extract key waypoints. Finally, the same spline curve or fifth-order polynomial interpolation algorithm is used to generate a new local cruise trajectory that satisfies flight control dynamics constraints (continuous velocity and acceleration).

[0029] Finally, the new local cruise trajectory is seamlessly integrated into the currently executing global cruise trajectory. The entire process does not require hovering or waiting, thus achieving continuous and uninterrupted execution of the mission.

[0030] Key constraints: The local replanning process only applies to obstacles that dynamically appear within non-no-fly zones; throughout the entire mission, the no-fly zone set issued during the initialization phase remains absolutely static, and its boundaries and extent cannot be modified or covered by any sensing data; if the no-fly zone settings need to be updated, they must be reconfigured and issued by the ground control terminal during the initialization of the next mission after the mission is completely completed.

[0031] Example 2 like Figure 2 As shown, Embodiment 2 of the present invention provides an autonomous cruise and obstacle avoidance system for unmanned aerial vehicles (UAVs) to implement the method described in Embodiment 1; the system includes two main parts: a ground control terminal and an UAV onboard system, which interact with each other through a communication module.

[0032] The ground control unit is responsible for generating and issuing a set of no-fly zones to the UAV's onboard system during the mission initialization phase; and for receiving and monitoring the status information, location, and map data transmitted back by the UAV during the UAV mission.

[0033] The UAV onboard system, integrated into the UAV platform, includes the following modules: The perception module synchronously acquires environmental images and its own attitude data through visual sensors (such as cameras) and an inertial measurement unit (IMU). In a preferred embodiment, the perception module also integrates an edge behavior recognition unit. This unit performs real-time analysis of the visual data stream based on onboard computing resources, and can identify specific target behaviors such as crowds and abnormal vehicle movements. The recognition results are then fed back as dynamic information to the path planning module for real-time adjustment of patrol point priorities.

[0034] The hierarchical SLAM mapping module is connected to the perception module, receives the raw data collected by the perception module, and runs the hierarchical SLAM mapping process as described in Example 1. This module performs real-time pose estimation through a tightly coupled visual inertial odometry (VIO) and continuously builds, fuses, and updates an occupied grid map containing no-fly zone markings.

[0035] The path planning module, connected to the hierarchical SLAM mapping module, executes the global patrol path planning process as described in Example 1 based on a real-time updated occupied grid map, generating an initial cruise trajectory covering the mission area. This module has an event-triggered mechanism; when the hierarchical SLAM mapping module detects a sudden obstacle in a non-no-fly zone, this mechanism is activated, and the control module then executes a local replanning process, generating a new local cruise trajectory through local path calculation, stitching, and optimization.

[0036] The trajectory optimization module, connected to the path planning module, is responsible for executing the trajectory generation process as described in Example 1. It compresses, samples key points, and performs smooth interpolation on the trajectory output by the path planning module, ultimately generating the final cruise trajectory that meets the speed and acceleration continuity requirements of the flight control system.

[0037] The flight control execution module, connected to the trajectory optimization module, receives and executes the final cruise trajectory (including the global cruise trajectory and the trajectory segments issued by local replanning), controls the UAV to complete high-precision trajectory tracking and attitude stabilization, and achieves smooth and seamless connection between different trajectory segments.

[0038] The communication module, serving as the data transmission channel between the UAV and the ground control terminal, is responsible for transmitting no-fly zone configurations, flight status, control commands, and updated map data. This communication module includes a redundant communication unit that supports dual-path redundant switching between low-latency short-range wireless links (such as data radios) and public cellular links (such as 4G / 5G) to ensure reliable transmission of mission data and map updates in complex electromagnetic environments or areas with signal obstruction.

[0039] System workflow: Before the mission begins, the ground control terminal sends the no-fly zone data set to the airborne system via the communication module. During mission execution, the perception module and the hierarchical SLAM mapping module work continuously to complete environmental perception and map construction; the path planning module performs global planning based on this data, and the trajectory optimization module smooths and optimizes the planned path before handing it over to the flight control execution module for cruise mission execution. During flight, if the system detects a sudden obstacle in a non-no-fly zone, the event triggering mechanism in the path planning module will be activated, quickly generating a local obstacle avoidance trajectory. This trajectory is then processed by the trajectory optimization module and seamlessly integrated by the flight control execution module for continued execution. Throughout the entire process, the system status is transmitted back to the ground control terminal in real time via the communication module. Through the coordinated work of the above modules, this system can achieve fully autonomous, high-coverage cruise and online intelligent obstacle avoidance of the UAV while strictly adhering to the constraints of static no-fly zones.

[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An unmanned aerial vehicle autonomous cruising and obstacle avoidance method for dynamic flight restricted area, characterized in that, The method comprises the following steps: S1, task initialization and static no-fly zone setting: in the task initialization stage, a no-fly zone set is generated by a ground control terminal and issued to a UAV system; the no-fly zone set remains static during the entire duration of the UAV flight task and does not receive any modification or update; S2, environment perception and hierarchical SLAM mapping: environment data is collected by a visual sensor and an inertial measurement unit on the UAV and input to a hierarchical SLAM mapping module for tight coupling processing, real-time estimation of the UAV pose and construction and update of an occupancy grid map, wherein the grid corresponding to the no-fly zone set is permanently marked as an impassable state in the occupancy grid map; S3, gridding modeling and reachable point set generation: based on the occupancy grid map, the patrol area is discretized into regular grids, the grids occupied by obstacles and the grids permanently marked as impassable are removed, and all passable grid points are extracted to form a reachable point set; S4, global patrol path planning: based on the passable grid point set, global path planning is performed, which includes: S41, BFS local shortest path search: an improved breadth-first search algorithm is used to calculate the shortest feasible path for any two passable grid points in the reachable point set; S42, global patrol sequence generation: taking the task starting point as the initial point, the next patrol point is selected in turn using a weighted greedy strategy to generate a global access sequence covering the reachable point set; S43, path splicing: the local shortest path segments between adjacent points in the global access sequence are sequentially spliced to form an initial cruise trajectory; S44, trajectory generation: the initial cruise trajectory is compressed and smoothed, including deleting consecutive intermediate nodes with the same direction, sampling an equal interval for a straight line segment with a length exceeding a threshold to retain key waypoints, and using a smoothing interpolation algorithm to generate a final cruise trajectory meeting the continuity requirements of the flight control system speed and acceleration; S5, trajectory issuance and autonomous cruise: the final cruise trajectory is issued to the flight control system to control the UAV to perform an autonomous cruise task; S6, dynamic obstacle avoidance and local trajectory re-planning: during the execution of the autonomous cruise, if the hierarchical SLAM mapping module detects a sudden obstacle in a non-no-fly zone and affects the currently executed trajectory, local trajectory re-planning is triggered; based on the current UAV position and obstacle information, a passable waypoint is reselected in a local range and a new local cruise trajectory is generated; the new local cruise trajectory is seamlessly connected to the global cruise trajectory being executed, while ensuring that the no-fly zone set remains unchanged. 2.The method of claim 1, wherein, The process of generating a new local cruise trajectory includes: S61, local path calculation: based on the reselected passable waypoint, the improved breadth-first search algorithm is used to calculate the local shortest feasible path from the current UAV position to the new waypoint and between the new waypoints; S62, local path splicing: the calculated local shortest feasible paths are sequentially spliced to form a local initial path; S63, local trajectory optimization: compress and smooth the local initial path to generate the new local cruise trajectory. 3.The method of claim 1, wherein, The execution process of the layered SLAM mapping module includes: extract visual features from image frames and time-align with IMU data through the feature extraction layer; construct a local occupancy grid map based on the aligned data through the local mapping layer, and submit local constraints to the global optimization layer; through the global optimization layer, globally correct the UAV pose and map based on loop detection and tight coupling optimization to reduce cumulative error. 4.The method of claim 1, wherein, The improved breadth-first search algorithm specifically includes: Before searching the path between the starting point and the target point, the connectivity of the reachable point set is analyzed, and the set is divided into several connected subgraphs; determine whether the starting point and the target point belong to the same connected subgraph; if not, evaluate the path reachability based on the precomputed boundary connection cost table or trigger cross-subgraph merging search; if they belong to the same subgraph, execute the standard BFS algorithm in the four-neighbor direction, and output the shortest path sequence.

5. The method of claim 1, wherein, In the weighted greedy strategy, the comprehensive weight function w(p) for selecting the next patrol point is defined as: ； where S k is the current location, p is the candidate target point, dist(S k , p) is the Manhattan distance between the current location and the candidate target point, q (p) is the importance coefficient of the candidate target point based on the task demand, a (p) is the accessibility score calculated based on the historical path state and the access frequency, and α, β, and γ are preset adjustable non-negative weight coefficients and satisfy α + β + γ = 1. 6.The method of claim 1, wherein, The smoothing interpolation algorithm uses a spline curve or a quintic polynomial to generate a smooth trajectory that meets the continuity requirements of the UAV flight control system speed and acceleration.

7. An unmanned aerial vehicle autonomous cruising and obstacle avoidance system for implementing the method of any one of claims 1-6, characterized in that, It includes: a ground control terminal for issuing the no-fly zone set during the task initialization stage and receiving the state information of the UAV during the task; a perception module for collecting environmental data through visual sensors and inertial measurement units; a layered SLAM mapping module for running the layered SLAM mapping steps to generate and maintain the occupancy grid map; a path planning module for executing the global cruise path planning steps to generate the initial cruise trajectory; a trajectory optimization module for executing the trajectory generation steps to generate the final cruise trajectory; a flight control execution module for receiving and executing the final cruise trajectory to control the UAV to complete the cruise task; a communication module for transmitting data between the UAV and the ground control terminal; a ground control terminal for issuing the no-fly zone set during the task initialization stage and receiving the state information of the UAV during the task. 8.The UAV autonomous cruising and obstacle avoidance system for dynamic flight restricted zone according to claim 7, wherein, The communication module includes a redundant communication unit that supports dual-path redundant switching of low-latency short-range wireless links and public cellular links to ensure reliable transmission of task data and updated maps in complex environments. 9.The UAV autonomous cruising and obstacle avoidance system for dynamic no-fly zone according to claim 7, wherein, The perception module further integrates an edge behavior recognition unit that performs real-time analysis of visual data streams based on onboard computing resources to identify specific target behaviors and feed the recognition results as dynamic inputs to the target importance coefficient q to the path planning module for real-time adjustment of patrol priority. 10.The UAV autonomous cruising and obstacle avoidance system for dynamic no-fly zone according to claim 7, wherein, The path planning module has an event triggering mechanism for generating a new local cruise trajectory by performing local path calculation, local path splicing, and local trajectory optimization when the dynamic obstacle avoidance and local trajectory re-planning are triggered, and preferentially issuing it to the flight control execution module for seamless trajectory access.

Images