Unmanned aerial vehicle autonomous obstacle avoidance method and system based on A star algorithm

Abstract

The invention provides an unmanned aerial vehicle autonomous obstacle avoidance method and system based on an A star algorithm, belongs to the technical field of unmanned aerial vehicle flight control, and solves the problems of path planning and autonomous obstacle avoidance of an unmanned aerial vehicle in a complex environment. 2, performing unmanned aerial vehicle path planning in the constructed environment map based on an A star algorithm; 3, constructing a simulation platform, and verifying the A star algorithm planning path by using the constructed simulation platform; and step 4, evaluating a verification result, and outputting an unmanned aerial vehicle planning path conforming to evaluation parameters.

Description

Technical Field

[0001] The invention relates to an autonomous obstacle avoidance method and system for an unmanned aerial vehicle based on an A-star algorithm, and belongs to the technical field of unmanned aerial vehicle flight control. Background Art

[0002] UAV technology has developed rapidly in recent years and is widely used in aerial photography, logistics, disaster relief, and environmental monitoring. However, with the increasing demand for UAVs to perform autonomous tasks in complex environments, autonomous obstacle avoidance and path planning technologies have become key factors restricting the development of UAVs. The core of autonomous flight of UAVs in complex terrain lies in the path planning algorithm. Designing an efficient path planning algorithm is the key to ensuring the smooth completion of UAV missions. Traditional path planning algorithms still have some bottlenecks when dealing with complex environments, especially when UAVs face dynamic obstacles or unknown environments. How to improve path planning efficiency while ensuring safety has become a research hotspot.

[0003] Among the existing path planning algorithms, the A-star algorithm is widely used in path planning in known environments due to its high search efficiency and path optimality. The A-star algorithm can effectively reduce the number of nodes during the search and find the optimal path to the target through the comprehensive calculation of node costs and heuristic functions. The A-star algorithm is usually applied to raster maps, especially in scenarios where the map is static and known. However, in complex environments, the A-star algorithm lacks path smoothness and has weak real-time performance. In addition, the path search efficiency of the A-star algorithm decreases as the complexity of the environment increases, limiting its application scenarios in dynamic or complex environments.

[0004] The existing path planning algorithms have the following deficiencies: 1. Efficiency bottleneck of the algorithm in complex environments: The current path planning algorithms often take a long time to calculate when facing highly complex environments. Although the A-star algorithm has a good path optimization effect on static maps, the path planning process consumes more computing resources when the environmental obstacles are complex and the path requirements are fine, which makes it difficult to meet the needs of drones in real-time tasks. 2. Insufficient path smoothness and obstacle avoidance accuracy: The paths generated by the A-star algorithm are often relatively discrete, especially when searching for paths in grid maps, the path smoothness is poor, which makes it more difficult for drones to follow the path in actual flight. In addition, the A-star algorithm lacks the ability to adapt to dynamic and static obstacles, which affects the actual obstacle avoidance effect of drones in complex environments and increases the uncertainty of flight. 3. Lack of simulation verification in complex environments: Although the A-star algorithm has a higher path planning performance in theory, it has less experimental verification in a high-simulation environment. Some existing technologies rely only on theoretical derivation and simplified experimental scenarios for verification, and fail to fully evaluate their path planning effects in scenarios with real complexity and obstacle interference, which leads to poor reliability and practicality of path planning in practical applications. Summary of the invention

[0005] The present invention solves the path planning and autonomous obstacle avoidance problems of UAVs in complex environments, and further proposes a method and system for autonomous obstacle avoidance of UAVs based on the A-star algorithm. The method for autonomous obstacle avoidance of UAVs based on the A-star algorithm includes the following steps:

[0006] Step 1: Build an environment map;

[0007] Step 2: Plan the UAV path in the constructed environment map based on the A-star algorithm;

[0008] Step 3: Build a simulation platform and use the built simulation platform to verify the path planned by the A-star algorithm;

[0009] Step 4: Evaluate the verification results and output the UAV planning path that meets the evaluation parameters.

[0010] Preferably, step 1 specifically includes:

[0011] Step 1.1: Set obstacles of different heights, positions and shapes in the Gazebo platform, build a three-dimensional simulation environment containing multiple obstacles, and create a grid map through the SLAM algorithm, where the SLAM algorithm includes but is not limited to the Gmapping algorithm, the Cartographer algorithm and the Hector algorithm;

[0012] Step 1.2: Use the RViz module in the ROS system to visualize the environment map built by the Gazebo platform, and mark the obstacle areas and passable areas in the two-dimensional map to complete the construction of the environment map.

[0013] Preferably, step 2 specifically includes:

[0014] Step 2.1: Take each grid cell in the two-dimensional map as a node of the A-star algorithm, where each node has location information, node cost and parent node information;

[0015] Step 2.2: Initialize the starting node and the target node, calculate the cost of each node through the cost function, and select the node with the smallest cost as the starting node;

[0016] Step 2.3: Traverse the adjacent nodes of the starting node and calculate the cost of each adjacent node;

[0017] Step 2.4: Update the cost and parent node information of the adjacent nodes, select the adjacent node with the smallest cost as the next forward node, store the explored nodes in the closed list, and store the unexplored nodes in the open list;

[0018] Step 2.5: Repeat steps 2.2 to 2.4 until the target node is found or the open list is empty, completing the path exploration, recording the complete path from the start node to the target node, and taking the complete path from the start node to the target node as the UAV planning path;

[0019] Step 2.6: Perform interpolation and smoothing of the path nodes in the planned path of the UAV to optimize the planned path of the UAV;

[0020] The expression of the cost function is:

[0021] F(n)=G(n)+H(n)(1);

[0022] In formula (1), G(n) is the actual cost from the starting node to node n, and H(n) is the heuristic estimate, that is, the estimated cost from node n to the target node. The H value is estimated using the Euclidean distance.

[0023] Preferably, step 3 specifically includes:

[0024] Step 3.1: Use the ROS system and Gazebo platform to build a drone simulation experiment platform;

[0025] Step 3.2: Load the 3D environment containing multiple obstacles in the simulation environment of the Gazebo platform, publish the relevant environment information through the ROS system, repeat step 2, and generate the corresponding 3D environment UAV planning path;

[0026] Step 3.3: Publish the planned path of the drone in the corresponding three-dimensional environment to the Gazebo platform, and use the drone's path following controller to make the drone fly along the planned path.

[0027] Preferably, step 4 specifically includes:

[0028] Step 4.1: Analyze the deviation between the actual flight path of the UAV in the UAV simulation experiment and the planned path of the UAV generated by the A-star algorithm, and evaluate the accuracy of the path;

[0029] Step 4.2: Analyze the number and angle of turning points in the A-star algorithm planning path, and evaluate whether the UAV planning path generated by the A-star algorithm meets the stability requirements of UAV flight;

[0030] Step 4.3: Count the number of times the drone avoids obstacles and reaches the target point during the flight, and evaluate the obstacle avoidance effect of the drone planning path generated by the A-star algorithm in complex environments;

[0031] Step 4.4: If the UAV planning path generated by the A-star algorithm does not meet any of the evaluation conditions in steps 4.1-4.3, reselect the starting point node and use the A-star algorithm to generate a new UAV planning path until the generated new UAV planning path meets any of the evaluation conditions in steps 4.1-4.3.

[0032] An autonomous obstacle avoidance system for UAV based on A-star algorithm, comprising:

[0033] An environment modeling module, a path planning module, and a path publishing and path following module connected in sequence;

[0034] The environment modeling module contains mapping functions for marking obstacles and traversable areas, which is used to generate two-dimensional maps based on raster maps;

[0035] The path planning module is based on the A-star algorithm, which is used to calculate the optimal path of the drone from the starting point to the target point in the two-dimensional map;

[0036] The path publishing and path following modules are used to publish the optimal path information generated by the path planning module through the ROS system and control the UAV to fly along the optimal path.

[0037] The beneficial effects of the present invention are:

[0038] 1. The path planning efficiency is high and meets the real-time requirements. The present invention adopts the A-star algorithm for global path planning, models the path planning problem of the UAV as a graph search problem, and enables the algorithm to quickly find the optimal path by reasonably designing the cost function and heuristic function. The A-star algorithm has a high search efficiency on static known maps and can achieve fast path calculation in complex environments. Compared with the Dijkstra algorithm, the heuristic estimation of the A-star algorithm reduces the search space and significantly improves the calculation speed, thereby meeting the UAV's demand for real-time path planning in complex mission scenarios, and effectively improving the execution efficiency of the UAV and the reliability of task completion.

[0039] 2. The path planning is accurate and the obstacle avoidance effect is significant. The present invention uses the A-star algorithm to search for paths on a grid map, and combines the precise distribution information of obstacles in the environment to plan an accurate and feasible path for the drone. The A-star algorithm can effectively avoid obstacle areas during the path planning process and achieve good obstacle avoidance effects. Compared with other path planning algorithms with strong randomness, the A-star algorithm has the advantages of high path controllability and stable results. It can ensure the continuity and feasibility of the path in a known obstacle environment, effectively improving the obstacle avoidance accuracy and success rate of drones in complex environments, and enhancing the reliability of path planning.

[0040] 3. Based on the verification of the high-simulation environment, the feasibility and practicality of path planning are improved. The present invention is verified in the high-simulation environment of ROS and Gazebo, and a complex environment containing multiple obstacles is built to simulate the path planning and obstacle avoidance effects in real UAV mission scenarios. Through simulation experiments, the present invention verifies the accuracy and effectiveness of path planning, and ensures the feasibility and reliability of the algorithm solution in practical applications. Compared with algorithms that lack simulation verification, the present invention is supported by sufficient experimental data, making it more applicable in real application scenarios, and can provide an effective path planning solution for autonomous flight of UAVs.

[0041] 4. Low implementation cost and simple structure. The present invention adopts the mature A-star algorithm and standard grid map modeling method, avoiding complex modifications to the algorithm structure and additional hardware requirements. The algorithm is simple to implement and easy to run on existing computing platforms. Using the open source ROS and Gazebo simulation environment, the present invention constructs a low-cost simulation verification platform, avoiding the high cost of actual experiments. Compared with path planning solutions that rely on dedicated hardware, the present invention not only reduces the implementation cost, but also makes the deployment of path planning more efficient, with good economy and operability.

[0042] 5. Through the above improvements, the present invention has significant advantages in path planning efficiency, obstacle avoidance capability, reliability of experimental verification and implementation cost, and provides a practical solution for autonomous path planning of UAVs in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of a flow chart of an autonomous obstacle avoidance method for a UAV based on the A-star algorithm provided by the present invention;

[0044] Figure 2 A schematic diagram of the environment modeling provided by the present invention, Figure 2 In the figure, (a) is a top view of the environment modeling including obstacles, and (b) is a side view of the environment modeling including obstacles;

[0045] Figure 3 This is a diagram showing the path following effect of a drone in a Gazebo simulation environment in an obstacle-free environment provided by the present invention. Figure 3 In the figure, (a) is a schematic diagram of path planning in an obstacle-free environment, and (b) is a schematic diagram of the position of the UAV;

[0046] Figure 4 This is a diagram of the path following effect of a drone in a Gazebo simulation environment in an environment with obstacles provided by the present invention. Figure 4 In the figure, (a) is a schematic diagram of path planning in an environment with complex obstacles, and (b) is a schematic diagram of the position of the UAV;

[0047] Figure 5 A schematic diagram of the flow chart of the A-star algorithm provided by the present invention;

[0048] Figure 6 A schematic diagram of the structure of an autonomous obstacle avoidance system for a UAV based on the A-star algorithm provided by the present invention. DETAILED DESCRIPTION

[0049] Specific implementation method 1: Combination Figure 1-5 To illustrate this embodiment, Figure 1 As shown, the steps of the autonomous obstacle avoidance method for a drone based on the A-star algorithm described in this embodiment include:

[0050] S1: Build an environment map;

[0051] S101: Use the ROS system and the Gazebo platform environment to build a three-dimensional simulation environment for the drone. The constructed environment contains obstacles and complex terrain, which is used to simulate the path planning scenario in the actual operation of the drone. The specific steps are as follows:

[0052] S102: Create a simulation scene in the Gazebo platform, set obstacles of different heights, positions and shapes, and build a complex environment with multiple obstacles to verify the obstacle avoidance effect of the path planning algorithm. The constructed environment map is as follows: Figure 2 (a) and Figure 2 (b) as shown.

[0053] S2: UAV path planning in the constructed environment map based on the A-star algorithm;

[0054] In this implementation, each grid unit in the two-dimensional map is used as a node of the A-star algorithm, wherein each node has position information, node cost and parent node information;

[0055] The expression of the cost function is:

[0056] F(n)=G(n)+H(n)(1);

[0057] In formula (1), G(n) is the actual cost from the starting node to node n, H(n) is the heuristic estimate, that is, the estimated cost from node n to the target node, and the H value is estimated using the Euclidean distance.

[0058] like Figure 5 As shown, the A-star algorithm includes:

[0059] S201: Select a node with the minimum cost from the Open List as the starting point node;

[0060] S202: remove the current node from the Open List and add it to the Close List, where the Open List is an open list for storing nodes to be explored, and the Close List is a closed list for storing nodes that have been explored;

[0061] S203: Traverse the adjacent nodes of the current node and calculate the cost of each adjacent node;

[0062] S204: Update the cost and parent node information of the adjacent node and add it to the Open list;

[0063] S205: Repeat S201-S204 until the target node is found or the Open List is empty.

[0064] This implementation uses the A-star algorithm to search for paths on a grid map, and combines the precise distribution information of obstacles in the environment to plan an accurate and feasible path for the drone. The A-star algorithm can effectively avoid obstacle areas during the path planning process and achieve good obstacle avoidance effects. Compared with other path planning algorithms with strong randomness, the A-star algorithm has the advantages of high path controllability and stable results. It can ensure the continuity and feasibility of the path in a known obstacle environment, effectively improve the obstacle avoidance accuracy and success rate of drones in complex environments, and enhance the reliability of path planning.

[0065] S3: Build a simulation platform and use it to verify the path planned by the A-star algorithm;

[0066] The environment for constructing the simulation platform in this implementation is as follows: the ROS system version is ROS noetic version, the Linux operating system is Ubuntu 20.04, and the Gazebo platform version is version 11.0.

[0067] ROS (Robot Operating System) is an open source robot development framework. It is a system that can provide a flexible and scalable software platform for robots. Although ROS is called an "operating system", it is actually a software framework running on the Linux operating system. ROS adopts a loosely coupled communication mode between nodes, where nodes are the basic operating units in the system and communicate through topics, services, and parameter servers. In ROS (Robot Operating System), a topic is a message passing mechanism that allows nodes to share and transmit data by publishing and subscribing to specific topics. Nodes can publish messages to a topic, and other nodes can receive corresponding messages by subscribing to the topic to achieve communication. Service is a communication method for request and response between nodes; a node can provide a service, and other nodes can call this service to request specific functions or data. The mechanism of service is similar to remote procedure call (RPC), which is very useful when immediate response is required; parameter server is a distributed storage system used to store and manage configuration parameters in ROS. Nodes can read parameters from parameter servers or write parameters into them, thereby realizing dynamic configuration and management of parameters. This mechanism enables the system to flexibly adjust parameter settings, improving the configurability and maintainability of the system. In addition, ROS also uses the concept of packages to store code. Each package contains the source code, compilation configuration files and other resources of related functions. Developers can selectively use and customize these packages as needed.

[0068] Gazebo is a robot simulator that is widely used in the research, development, and testing of robotics. As a highly flexible and powerful tool, Gazebo is able to simulate a variety of robots, sensors, and environments, providing a simulation environment so that developers can test their algorithms and systems in a virtual environment without the need for actual hardware. Gazebo has a highly realistic physics engine, ROS interface, sensor simulation, etc. Gazebo uses a physics-based simulation engine that can simulate real-world physical phenomena such as gravity, friction, collision, etc., thereby achieving a highly realistic simulation of the robot. This enables developers to test their control algorithms and path planners in a virtual environment, as well as simulate various tasks for the robot. Gazebo provides an extensible ROS interface that enables it to be integrated with ROS. Through the ROS interface, developers can use various tools and libraries in ROS to control and monitor the simulation environment in Gazebo, thereby achieving comprehensive simulation and testing of the robot system. This integrated feature makes Gazebo an important part of the ROS ecosystem, providing a convenient simulation tool for robot developers; Gazebo has rich sensor simulation functions, including lidar, camera, etc. Developers can simulate the output of various sensors in the simulation environment to test and evaluate the performance of the robot system in different environments and conditions. This sensor simulation function provides a convenient tool for robot developers to help them better understand and debug the robot system; Gazebo also has a flexible environment editing function, allowing developers to create and edit various virtual environments. Developers can add terrain, buildings, obstacles and other elements in Gazebo to simulate the environmental scenes in this embodiment for testing and evaluating the performance of the robot system. This flexible environment editing function makes Gazebo a powerful simulation tool, providing robot developers with a laboratory for studying and testing various robot algorithms and systems; Gazebo is a powerful, flexible and easy-to-use simulator that provides a realistic virtual environment for the research, development and testing of robot technology. Its rich features make it a simulation tool for robot developers, helping developers to better understand, debug and optimize the performance of the robot system. By adding obstacles, terrain and other elements to Gazebo, the simulation environment simulation of drone path planning can be realized.

[0069] Rviz (ROS Visualization) is a 3D visualization tool in ROS. Its function is to provide robot developers with an interactive interface for visualizing various data and states in the robot system. As part of the ROS system, Rviz plays a vital role. It can be seamlessly integrated with other nodes and packages in ROS to achieve comprehensive visualization of the robot system. Through Rviz, you can observe the robot's behavior, perceived environment, tasks performed, etc., so as to better understand the robot's working principle and performance. The main functions of Rviz include sensor data visualization, path planning result display, scene construction, etc. Rviz can visualize various sensor data, such as lidar, camera, etc. Developers can view the data output by the sensor in real time, including point clouds, images, etc., in order to understand the situation of the robot's surrounding environment. In simulation experiments, Rviz is needed to obtain the environment map scanned by the SLAM lidar.

[0070] The simulation experiment in this embodiment includes an obstacle flight experiment and an obstacle-free flight experiment. Multiple obstacles are set in the environment map in S1, and the size of the environment map is set to 100*100.

[0071] In the obstacle-free flight experiment, Figure 3 As shown in (a), (0,0) is the coordinate of the starting node, (-18,-7) is the coordinate of the ending node, and the triangular area indicates the location of the drone. Figure 3 (b) is the location that the drone reaches after flying along the planned path, and the circle indicates the location of the drone. The flight speed of the drone is preset to the maximum horizontal speed of 0.3, and the horizontal height is set to 0.8. It can be seen that the drone can accurately reach the destination along the planned path without obstacles.

[0072] In the flight experiment with obstacles, Figure 4 As shown in (a), (0,0) is the starting node, (29, -10) is the end node coordinate, and the triangular area represents the location of the drone. Figure 4 (b) is the location that the UAV reaches after flying along the planned path. The circle indicates the location of the UAV. It can be seen that in an environment with complex obstacles, the UAV can accurately reach the destination along the path planned by the A-star algorithm.

[0073] S4: Evaluate the verification results and output the UAV planning path that meets the evaluation parameters;

[0074] During the experiment of S3, the effectiveness of the path planning and obstacle avoidance performance in this implementation were evaluated by recording the path execution of the UAV in the simulation environment. The following are the main parameters for the evaluation:

[0075] Path deviation: observe the deviation between the actual flight path of the drone and the path planned by the A-star algorithm, and evaluate the accuracy of the path;

[0076] Path smoothness: Analyze the number and angle of turning points in the path to evaluate whether the smoothness of the A-star algorithm path meets the stability requirements of the drone flight;

[0077] Obstacle avoidance success rate: Count the number of times the drone avoids obstacles and reaches the target point to evaluate the obstacle avoidance effect in complex environments;

[0078] It can be seen from the simulation experiment results that, given a simulation map and on the premise of autonomously setting the starting point and end point, the UAV can fly along the path planned by the A* algorithm, successfully avoid obstacles, and reach the set end point. This shows that the path planning algorithm designed in this embodiment is accurate and practical. Compared with algorithms that lack simulation verification, the present invention is supported by sufficient experimental data, making it more applicable in real application scenarios and providing an effective path planning solution for autonomous flight of UAVs.

[0079] In summary, this implementation adopts the mature A-star algorithm and the standard grid map modeling method, avoiding the complex modification of the algorithm structure and the requirement for additional hardware. The algorithm is simple to implement and easy to run on the existing computing platform. By using the open source ROS and Gazebo simulation environment, the present invention constructs a low-cost simulation verification platform, avoiding the high cost investment of actual experiments. Compared with the path planning scheme that relies on dedicated hardware, this implementation not only reduces the implementation cost, but also makes the deployment of path planning more efficient, with good economy and operability.

[0080] Specific implementation method 2: Combination Figure 6 This embodiment is described as follows. Figure 6 As shown, the autonomous obstacle avoidance system for unmanned aerial vehicles based on the A-star algorithm described in this embodiment includes: an environment modeling module, a path planning module, and a path publishing and path following module connected in sequence;

[0081] The environment modeling module contains mapping functions for marking obstacles and traversable areas, which is used to generate two-dimensional maps based on raster maps;

[0082] The path planning module is based on the A-star algorithm, which is used to calculate the optimal path of the drone from the starting point to the target point in the two-dimensional map;

[0083] The path publishing and path following modules are used to publish the optimal path information generated by the path planning module through the ROS system and control the UAV to fly along the optimal path.

[0084] The above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent replacement and improvement made to the above embodiments without departing from the content of the technical solution of the present invention, based on the technical essence of the present invention, within the spirit and principles of the present invention, still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for autonomous obstacle avoidance of a UAV based on the A-star algorithm, characterized in that: The steps of the autonomous obstacle avoidance method for unmanned aerial vehicles based on the A-star algorithm include: Step 1: Build an environment map; Step 2: Plan the UAV path in the constructed environment map based on the A-star algorithm; Step 3: Build a simulation platform and use the built simulation platform to verify the path planned by the A-star algorithm; Step 4: Evaluate the verification results and output the UAV planning path that meets the evaluation parameters.

2. The autonomous obstacle avoidance method for a UAV based on the A-star algorithm according to claim 1 is characterized in that: Step 1 specifically includes: Step 1.1: Set obstacles of different heights, positions and shapes in the Gazebo platform, build a three-dimensional simulation environment containing multiple obstacles, and create a grid map through the SLAM algorithm, where the SLAM algorithm includes but is not limited to the Gmapping algorithm, the Cartographer algorithm and the Hector algorithm; Step 1.2: Use the RViz module in the ROS system to visualize the environment map built by the Gazebo platform, and mark the obstacle areas and passable areas in the two-dimensional map to complete the construction of the environment map.

3. The autonomous obstacle avoidance method for a UAV based on the A-star algorithm according to claim 1 is characterized in that: Step 2 specifically includes: Step 2.1: Take each grid cell in the two-dimensional map as a node of the A-star algorithm, where each node has location information, node cost and parent node information; Step 2.2: Initialize the starting node and the target node, calculate the cost of each node through the cost function, and select the node with the smallest cost as the starting node; Step 2.3: Traverse the adjacent nodes of the starting node and calculate the cost of each adjacent node; Step 2.4: Update the cost and parent node information of the adjacent nodes, select the adjacent node with the smallest cost as the next forward node, store the explored nodes in the closed list, and store the unexplored nodes in the open list; Step 2.5: Repeat steps 2.2 to 2.4 until the target node is found or the open list is empty, completing the path exploration, recording the complete path from the start node to the target node, and taking the complete path from the start node to the target node as the UAV planning path; Step 2.6: Perform interpolation and smoothing of the path nodes in the planned path of the UAV to optimize the planned path of the UAV; The expression of the cost function is: F(n)=G(n)+H(n)(1); In formula (1), G(n) is the actual cost from the starting node to node n, H(n) is the heuristic estimate, that is, the estimated cost from node n to the target node, and the H value is estimated using the Euclidean distance.

4. The autonomous obstacle avoidance method for a UAV based on the A-star algorithm according to claim 1, characterized in that: Step 3 specifically includes: Step 3.1: Use the ROS system and Gazebo platform to build a drone simulation experiment platform; Step 3.2: Load the 3D environment containing multiple obstacles in the simulation environment of the Gazebo platform, publish the relevant environment information through the ROS system, repeat step 2, and generate the corresponding 3D environment UAV planning path; Step 3.3: Publish the planned path of the drone in the corresponding three-dimensional environment to the Gazebo platform, and use the drone's path following controller to make the drone fly along the planned path.

5. The autonomous obstacle avoidance method for a UAV based on the A-star algorithm according to claim 1, characterized in that: Step 4 specifically includes: Step 4.1: Analyze the deviation between the actual flight path of the UAV in the UAV simulation experiment and the planned path of the UAV generated by the A-star algorithm, and evaluate the accuracy of the path; Step 4.2: Analyze the number and angle of turning points in the A-star algorithm planning path, and evaluate whether the UAV planning path generated by the A-star algorithm meets the stability requirements of UAV flight; Step 4.3: Count the number of times the drone avoids obstacles and reaches the target point during the flight, and evaluate the obstacle avoidance effect of the drone planning path generated by the A-star algorithm in complex environments; Step 4.4: If the UAV planning path generated by the A-star algorithm does not meet any of the evaluation conditions in steps 4.1-4.3, reselect the starting point node and use the A-star algorithm to generate a new UAV planning path until the generated new UAV planning path meets any of the evaluation conditions in steps 4.1-4.

3.

6. An autonomous obstacle avoidance system for unmanned aerial vehicles based on the A-star algorithm, applied to an autonomous obstacle avoidance method for unmanned aerial vehicles based on the A-star algorithm as claimed in any one of claims 1 to 5, characterized in that: include: An environment modeling module, a path planning module, and a path publishing and path following module connected in sequence; The environment modeling module includes a mapping function for marking obstacles and traversable areas, and is used to generate a two-dimensional map based on a grid map; The path planning module is based on the A-star algorithm, which is used to calculate the optimal path of the drone from the starting point to the target point in the two-dimensional map; The path publishing and path following module is used to publish the optimal path information generated by the path planning module through the ROS system, and control the UAV to fly along the optimal path.

Images