Multi-modal information collaborative all-dimensional autonomous obstacle avoidance algorithm for unmanned aerial vehicle

Abstract

The invention provides a multi-modal information collaborative unmanned aerial vehicle omnibearing autonomous obstacle avoidance algorithm, and the algorithm comprises a data collection module, a data processing module, and a path planning and obstacle avoidance module, and is used for carrying out the unmanned aerial vehicle omnibearing autonomous obstacle avoidance according to a processed result. An improved artificial potential field method and a sparrow search algorithm are utilized to plan a flight path of the unmanned aerial vehicle, repulsive force is calculated, and a target approaching factor is introduced, so that when the unmanned aerial vehicle approaches a target, the repulsive force is reduced, autonomous obstacle avoidance is completed, and an execution module and a communication and monitoring module are controlled. According to the multi-modal information collaborative unmanned aerial vehicle omnibearing autonomous obstacle avoidance algorithm provided by the invention, secondary optimization is carried out on the path through a discoverer-follower labor division mechanism, a local optimal solution is effectively jumped out, a global optimal or approximately optimal path is generated, a shorter and safer path can be generated in a complex obstacle scene, and the detouring distance is reduced.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to an all-round autonomous obstacle avoidance algorithm for UAVs based on multimodal information collaboration. Background Technology

[0002] The ability of drones to autonomously avoid obstacles in complex and dynamic environments is a core technology to ensure their safe flight and mission efficiency. With the expansion of application scenarios, such as logistics delivery, disaster relief, and aerial monitoring, drones need to perceive obstacles in real time in unstructured environments, plan safe paths, and execute flight missions stably in order to avoid collisions, reduce risks, and improve operational efficiency.

[0003] However, traditional obstacle avoidance methods often rely on a single sensor, such as lidar, camera or simple rule control, which has problems such as limited perception range, poor adaptability and easy getting trapped in local optima, making it difficult to meet the omnidirectional autonomous obstacle avoidance requirements in high dynamic and multi-obstacle scenarios. Existing artificial potential field methods achieve path planning by constructing gravitational and repulsive fields, which have the advantages of high computational efficiency and strong real-time performance. However, when there are obstacles near the target point, the repulsive and gravitational forces will cancel each other out, causing the drone to oscillate or stop in front of the target point. Furthermore, in complex obstacle distributions, such as U-shaped obstacles or dense obstacle groups, the potential field is prone to forming local minima, causing the drone to get stuck in a deadlock state.

[0004] Therefore, it is necessary to provide a new multimodal information collaborative omnidirectional autonomous obstacle avoidance algorithm for UAVs to solve the above-mentioned technical problems. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a multimodal information collaborative omnidirectional autonomous obstacle avoidance algorithm for unmanned aerial vehicles (UAVs).

[0006] The multimodal information collaborative UAV all-round autonomous obstacle avoidance algorithm provided by the present invention includes: a data acquisition module, used to collect multimodal information during the flight of the UAV; The data processing module is used to analyze and process the collected multimodal information, identify obstacles and determine the position and distance of the obstacles relative to the UAV, and at the same time evaluate the flight status of the UAV. The path planning and obstacle avoidance module is used to plan the flight path of the UAV based on the processed results using the improved artificial potential field method and sparrow search algorithm, calculate the repulsive force, and introduce a target proximity factor so that the repulsive force is reduced when the UAV approaches the target, thus completing autonomous obstacle avoidance. The control execution module is used to adjust the drone's attitude and motor speed based on the path planning results, so as to enable the drone to fly stably and avoid obstacles. The communication and monitoring module is used to enable data transmission and communication between the UAV and the ground station.

[0007] Preferably, the flight path of the UAV is planned using the improved artificial potential field method and sparrow search algorithm. The specific operation steps are as follows: Step 1: Construct the initial potential field path: Based on the current position of the UAV, calculate the gravitational potential field and the improved repulsive potential field, superimpose the two to obtain the total potential field, and generate the initial path along the potential field gradient direction. Step 2: Initialize the sparrow population: Using the initial path as the center, randomly generate N candidate paths around the drone, which are the sparrow positions. Each path is represented by a discrete point, forming the initial position distribution of the sparrow population. Step 3: Calculate fitness value: After initializing the sparrow population, for each candidate path, comprehensively evaluate three indicators: minimum distance to obstacles, total path length, and rate of curvature change. Calculate the fitness value by weighted summation. Higher fitness indicates better path safety and efficiency. Step 4: Sparrow position update and global search: Based on the fitness value, the sparrow population is divided into discoverers and followers. The positions of discoverers and followers are updated to obtain the global optimal path. Step 5, Path Optimization and Output: When the number of iterations reaches the maximum value or the fitness value converges, the final globally optimal path is used as the flight path of the UAV, replacing the initial potential field path and output.

[0008] Preferably, the discoverer is responsible for global search and prioritizes exploring unvisited areas, while the followers are used for local development near the discoverer.

[0009] Preferably, in Step 2, the path point must meet the minimum safe distance constraint from the obstacle, that is, the distance from the current position of the UAV to the obstacle must be greater than the preset minimum safe distance, and the total path length must not exceed the preset threshold.

[0010] Preferably, the multimodal information includes visual images, ultrasonic distance data, and IMU (Inertial Measurement Unit) data. The visual sensor captures image information in front of the UAV, the ultrasonic sensor measures the distance between the UAV and the nearest obstacle, and the IMU integrates accelerometers, gyroscopes, and magnetometers to measure the attitude and motion state of the UAV, thus obtaining accelerometer, gyroscope, and magnetometer counts.

[0011] Preferably, the sparrow search algorithm includes the following steps: Step 1: Initialize the sparrow population: First, randomly generate a set of candidate solutions around the drone, that is, the positions of the sparrows. Each candidate solution represents a possible flight path. Step 2: Calculate the fitness value: For each candidate solution, calculate its fitness value, which is a comprehensive evaluation of the distance between the UAV and obstacles and the path length when flying along the path. Step 3: Update sparrow positions: According to the rules of the sparrow search algorithm, update the sparrow positions. The discoverer is responsible for the global search, and the follower is responsible for the local search. Through continuous iteration, the sparrow population gradually approaches the global optimum. Step 4: Path optimization: Use the globally optimal solution found by the sparrow search algorithm as the flight path of the UAV.

[0012] Preferably, the operation steps of the data processing module are as follows: S1. Visual Image Processing: First, visual image processing uses the YOLOv5 algorithm to identify obstacles and obtain their two-dimensional image coordinates. Combining the camera intrinsic parameter matrix and the distance value measured by the ultrasonic sensor, the image coordinates are converted into three-dimensional coordinates (x, y, z) in the UAV coordinate system, and the distance value is output. S2. Ultrasonic data processing: The output distance value is smoothed by moving average filtering technology, and the distance value measured by ultrasound is cross-validated by the results of visual image detection. When the distance deviation between visual and ultrasonic detection exceeds the preset threshold, data correction is triggered. S3, IMU data fusion: Complementary filtering is used to integrate accelerometer, gyroscope and magnetometer data to estimate the UAV's attitude angle, speed and altitude. At this time, the UAV motion state output by the IMU forms a spatial correlation with the obstacle distance data optimized in step S2, and the relative motion trend of the obstacle is predicted based on the UAV speed. S4. Status Assessment: By setting attitude angle thresholds and speed fluctuation ranges, the system determines whether the UAV is in a stable flight state. At the same time, the altitude data output by the IMU verifies the vertical accuracy of the ultrasonic ranging. The system integrates the obstacle's three-dimensional coordinates, distance, speed, and the UAV's own status parameters from steps S1-S3 into the input list required by the path planning and obstacle avoidance module.

[0013] Preferably, the attitude angles include the roll, pitch, and yaw of the UAV, the speed of the UAV is obtained by integrating accelerometer data, and the altitude data of the UAV is determined by fusing barometer data.

[0014] Preferably, the control execution module includes discretizing the path planning result into time series trajectory points, generating target attitude commands through a PID algorithm, calculating the differential speed of the motor and adding feedforward compensation, and simultaneously monitoring IMU data in real time. When the attitude angle deviation of the UAV exceeds a preset threshold, it automatically triggers the damping coefficient adjustment to maintain flight stability.

[0015] Compared with related technologies, the multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm provided by this invention has the following beneficial effects: Traditional artificial potential field methods are prone to causing drone oscillations or stagnation when there are obstacles near the target point due to the mutual cancellation of repulsive and attractive forces. This invention defines a new repulsive potential field function and introduces a distance attenuation factor and a target proximity factor, so that the repulsive force dynamically attenuates as the distance between the drone and the target decreases, avoiding the generation of a potential field equilibrium point, ensuring that the drone can smoothly reach the target point, avoiding oscillations or stagnation caused by the cancellation of repulsive and attractive forces, and improving the continuity of the obstacle avoidance path. In U-shaped obstacle or dense obstacle group scenarios, the traditional potential field method is prone to getting stuck in local minima. This invention combines the global search capability of the sparrow search algorithm, takes the initial path generated by the potential field method as the center, and performs secondary optimization of the path through the discoverer-follower division of labor mechanism, effectively escaping local optimal solutions and generating globally optimal or near-optimal paths. It can generate shorter and safer paths in complex obstacle scenarios and reduce detour distances. 3. Traditional methods rely on single sensors such as lidar and cameras, which have problems such as limited sensing range and poor anti-interference ability. This invention uses a data acquisition module to synchronously and collaboratively perceive the movement status of obstacles in real time through multimodal information collaboration. The data is transmitted to the ground station for remote monitoring and intervention through a communication module. Combined with the dynamic adjustment of repulsive force in the improved potential field method, the UAV can quickly respond to dynamic obstacles and avoid collisions with birds and other birds. Attached Figure Description

[0016] Figure 1 A system block diagram of the multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm provided by the present invention; Figure 2 This is a schematic diagram of the process structure for planning the flight path of a UAV using the improved artificial potential field method and sparrow search algorithm in this invention. Detailed Implementation

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] Please refer to the following: Figures 1 to 2 ,in, Figure 1 A system block diagram of the multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm provided by the present invention; Figure 2 This is a schematic diagram of the process structure for planning the flight path of a UAV using the improved artificial potential field method and sparrow search algorithm in this invention.

[0019] Example: PID algorithm, the control algorithm of PID controller. In process control, PID controller, also known as PID regulator, controls according to the proportional (P), integral (I), and derivative (D) of the deviation. It is the most widely used automatic controller. It has the advantages of simple principle, easy implementation, wide applicability, independent control parameters, and relatively simple parameter selection. Moreover, it can be theoretically proven that for typical process control objects with first-order lag + pure lag and second-order lag + pure lag, PID controller is an optimal control. PID regulation law is an effective method for dynamic quality correction of continuous systems. Its parameter tuning method is simple and its structure can be flexibly changed.

[0020] like Figure 1 As shown, the multimodal information collaborative UAV all-round autonomous obstacle avoidance algorithm includes a data acquisition module for collecting multimodal information during UAV flight. The multimodal information includes visual images, ultrasonic distance data, and IMU inertial measurement unit data. The visual sensor captures image information in front of the UAV, the ultrasonic sensor measures the distance between the UAV and the nearest obstacle, and the IMU inertial measurement unit integrates accelerometer, gyroscope and magnetometer to measure the attitude and motion state of the UAV, and obtains accelerometer, gyroscope and magnetometer data. The data processing module is used to analyze and process the collected multimodal information, identify obstacles and determine the position and distance of the obstacles relative to the UAV, and at the same time evaluate the flight status of the UAV. The operation steps of the data processing module are as follows: S1. Visual Image Processing: First, visual image processing uses the YOLOv5 algorithm to identify obstacles and obtain their two-dimensional image coordinates. Combining the camera intrinsic parameter matrix and the distance value measured by the ultrasonic sensor, the image coordinates are converted into three-dimensional coordinates (x, y, z) in the UAV coordinate system, and the distance value is output. S2. Ultrasonic data processing: The output distance value is smoothed by moving average filtering technology, and the distance value measured by ultrasound is cross-validated by the results of visual image detection. When the distance deviation between visual and ultrasonic detection exceeds the preset threshold, data correction is triggered. S3, IMU Data Fusion: Complementary filtering is used to fuse and integrate accelerometer, gyroscope and magnetometer data to estimate the UAV's attitude angle, speed and altitude. The attitude angle includes the UAV's roll, pitch and yaw. The UAV speed is obtained by integrating the accelerometer data. The UAV altitude data is determined by fusing barometer data. At this time, the UAV motion state output by the IMU forms a spatial correlation with the obstacle distance data optimized in step S2. The relative motion trend of the obstacle is predicted based on the UAV speed. S4. Status Assessment: By setting attitude angle thresholds and speed fluctuation ranges, the system determines whether the UAV is in a stable flight state. At the same time, the altitude data output by the IMU verifies the vertical accuracy of the ultrasonic ranging. The system integrates the obstacle's three-dimensional coordinates, distance, speed, and the UAV's own status parameters from steps S1-S3 into the input list required by the path planning and obstacle avoidance module.

[0021] The path planning and obstacle avoidance module is used to plan the flight path of the UAV based on the processed results using the improved artificial potential field method and sparrow search algorithm, calculate the repulsive force, and introduce a target proximity factor so that the repulsive force is reduced when the UAV approaches the target, thus completing autonomous obstacle avoidance. Improved artificial potential field methods include: Step 1: First, define the new repulsive potential field function: ; in, The repulsive potential field value from the obstacle experienced by the drone at its current position q; It is the distance from the drone's current position q to the obstacle; This is the preset minimum safe distance, i.e., the range of influence of the obstacle; It is the repulsive force gain coefficient; n is a positive number used to adjust the attenuation rate of the repulsive force. Step 2: After defining the parameters, calculate the repulsive force and introduce a target proximity factor so that the repulsive force decreases when the drone approaches the target. Specifically, this is implemented as follows:

[0022] in, Let q be the repulsive force exerted on the drone by the obstacle at its current position. It represents the repulsive force exerted by the obstacle on the drone, pointing away from the obstacle, and its magnitude is related to the distance between the drone and the obstacle. Represents the repulsive potential field function The gradient, i.e., the rate of change of the potential field at position q, points in the direction of the fastest growth of the potential field (i.e., away from the obstacle), and its negative value is: - That is, the direction of the repulsive force; It is the distance from the drone's current position q to the target point; It is the threshold distance; k is an adjustment factor used to control the rate of change of the target proximity factor.

[0023] The sparrow search algorithm includes the following steps: Step 1: Initialize the sparrow population: First, randomly generate a set of candidate solutions around the drone, that is, the positions of the sparrows. Each candidate solution represents a possible flight path. Step 2: Calculate the fitness value: For each candidate solution, calculate its fitness value, which is a comprehensive evaluation of the distance between the UAV and obstacles and the path length when flying along the path. Step 3: Update sparrow positions: According to the rules of the sparrow search algorithm, update the sparrow positions. The discoverer is responsible for the global search, and the follower is responsible for the local search. Through continuous iteration, the sparrow population gradually approaches the global optimum. Step 4: Path optimization: Use the globally optimal solution found by the sparrow search algorithm as the flight path of the UAV.

[0024] The control execution module is used to adjust the drone's attitude and motor speed based on the path planning results, so as to enable the drone to fly stably and avoid obstacles. The control execution module includes discretizing the path planning results into time series trajectory points, generating target attitude commands through a PID algorithm, calculating the differential speed of the motor and adding feedforward compensation, and monitoring IMU data in real time. When the attitude angle deviation of the UAV exceeds a preset threshold, it automatically triggers the damping coefficient adjustment to maintain flight stability. The communication and monitoring module is used to enable data transmission and communication between the UAV and the ground station.

[0025] It should be noted that, firstly, obstacle information, path planning results, and flight status parameters, including attitude, speed, and battery level, are transmitted to the ground station via Wi-Fi or 4G / 5G network modules. The drone receives instructions from the ground station, such as emergency hovering or returning to home. The ground station software, such as an application on a PC, tablet, or mobile phone, displays the drone's real-time location, obstacle distribution, planned path, and flight parameters, and supports 3D map overlay. Specifically, when there are obstacles near the target point, the traditional artificial potential field method is prone to causing the drone to oscillate or stop due to the mutual cancellation of repulsive and attractive forces. This invention defines a new repulsive potential field function and introduces a distance decay factor and a target proximity factor, so that the repulsive force dynamically decays as the distance between the drone and the target decreases, avoiding the generation of a potential field equilibrium point, ensuring that the drone can smoothly reach the target point, avoiding oscillation or stoppage caused by the cancellation of repulsive and attractive forces, and improving the continuity of the obstacle avoidance path. Furthermore, in U-shaped obstacle or dense obstacle group scenarios, the traditional potential field method is prone to getting stuck in local minima. This invention combines the global search capability of the sparrow search algorithm, takes the initial path generated by the potential field method as the center, and performs secondary optimization of the path through the discoverer-follower division of labor mechanism, effectively escaping local optimal solutions and generating globally optimal or near-optimal paths. It can generate shorter and safer paths in complex obstacle scenarios and reduce detour distances. Furthermore, traditional methods rely on single sensors such as lidar and cameras, which have problems such as limited sensing range and poor anti-interference ability. This invention uses a data acquisition module to synchronously and collaboratively perceive the movement status of obstacles in real time through multimodal information collaboration, and transmits the data to the ground station for remote monitoring and intervention through a communication module. Combined with the dynamic adjustment of repulsive force in the improved potential field method, the UAV can quickly respond to dynamic obstacles and avoid collisions with birds and other birds.

[0026] Example: Reference Figure 2 As shown, the improved artificial potential field method and sparrow search algorithm are used to plan the flight path of the UAV. The specific operation steps are as follows: Step 1: Construct the initial potential field path: First, calculate the gravitational potential field based on the current position q of the UAV. and improve the repulsive field At this point, the total potential field is: and along the direction of the potential field Generate initial path ; Step 2: Initialize the sparrow population: using the initial path Centered on the drone, N candidate paths, i.e., sparrow positions, are randomly generated around it, and each candidate path is represented by a discrete point. Among them, the path points must satisfy obstacle avoidance constraints, and the distance from the obstacles must be specified. And the length of the path points does not exceed the threshold. To prevent excessive energy consumption due to excessively long paths; Step 3: Calculate fitness values: After initializing the sparrow population, use the minimum distance between the path and obstacles. Calculate fitness based on path length L and path curvature change rate k. The calculation formula is: ; in, Weight coefficients, fitness It is a multi-objective evaluation function; Step 4: Sparrow position update: based on fitness Based on the evaluation results, we define the discoverer and followers, update their positions, and obtain the globally optimal path: Among them, the discoverer is responsible for global search and prioritizes exploration of unexplored areas; the follower is used to conduct local development near the discoverer; The formula for updating the discoverer's location is: ; This represents the updated position of the i-th sparrow at the j-th dimension or the j-th path point, which is the coordinate of the new path point obtained after iteration; i refers to the sparrow's number, ranging from... j represents the dimension of the path point. If the path is represented by discrete points, j corresponds to the index of the j-th path point; if it is a continuous space, j corresponds to the coordinate axis dimension. The step size control factor is T, where T is the maximum number of iterations. The globally optimal path; The follower position update formula is: ; As a random factor, This is the worst path; After each iteration, the fitness is re-evaluated, and the globally optimal path is updated. Step 5, Path Optimization and Output: When the maximum number of iterations T is reached or the fitness value converges, ... As the optimized path , replacement And output it.

[0027] Specifically, traditional methods are prone to getting trapped in local minima, such as in areas with dense obstacles, and the path may oscillate or pass through obstacles. By introducing a dynamic adjustment coefficient η to improve the repulsive field, the repulsive force changes more smoothly with distance, reducing local minima. Furthermore, by combining the global search capability of the sparrow search algorithm, it can escape local traps. Furthermore, safety can be considered simultaneously through the fitness function, such as... The path length L and the path curvature change rate k are used to avoid the defects caused by a single target, such as the shortest path possibly passing through obstacles or making sharp turns. The initial path is generated by using an artificial potential field method, that is, by constructing the gravitational potential field of the current position of the UAV and the improved repulsive potential field, the two are superimposed to generate the total potential field, and the initial path is generated along the gradient direction of the potential field. This provides a directional initial solution for the sparrow search, reduces the blindness of random search, and the discoverer step size decreases with the number of iterations, allowing for rapid exploration in the early stage and development in the later stage, thus improving the convergence efficiency.

[0028] The circuits and controls involved in this invention are all existing technologies and will not be described in detail here.

[0029] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A multimodal information-coordinated omnidirectional autonomous obstacle avoidance algorithm for unmanned aerial vehicles, characterized in that: Includes a data acquisition module for collecting multimodal information during the drone's flight; The data processing module is used to analyze and process the collected multimodal information, identify obstacles and determine the position and distance of the obstacles relative to the UAV, and at the same time evaluate the flight status of the UAV. The path planning and obstacle avoidance module is used to plan the flight path of the UAV based on the processed results using the improved artificial potential field method and sparrow search algorithm, calculate the repulsive force, and introduce a target proximity factor so that the repulsive force is reduced when the UAV approaches the target, thus completing autonomous obstacle avoidance. The control execution module is used to adjust the drone's attitude and motor speed based on the path planning results, so as to enable the drone to fly stably and avoid obstacles. The communication and monitoring module is used to enable data transmission and communication between the UAV and the ground station.

2. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 1, characterized in that, The flight path of the UAV is planned using an improved artificial potential field method and a sparrow search algorithm. The specific operation steps are as follows: Step 1: Construct the initial potential field path: Based on the current position of the UAV, calculate the gravitational potential field and the improved repulsive potential field, superimpose the two to obtain the total potential field, and generate the initial path along the potential field gradient direction. Step 2: Initialize the sparrow population: Using the initial path as the center, randomly generate N candidate paths around the drone, which are the sparrow positions. Each path is represented by a discrete point, forming the initial position distribution of the sparrow population. Step 3: Calculate fitness value: After initializing the sparrow population, for each candidate path, comprehensively evaluate three indicators: minimum distance to obstacles, total path length, and rate of curvature change. Calculate the fitness value by weighted summation. Higher fitness indicates better path safety and efficiency. Step 4: Sparrow position update and global search: Based on the fitness value, the sparrow population is divided into discoverers and followers. The positions of discoverers and followers are updated to obtain the global optimal path. Step 5, Path Optimization and Output: When the number of iterations reaches the maximum value or the fitness value converges, the final globally optimal path is used as the flight path of the UAV, replacing the initial potential field path and output.

3. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 2, characterized in that, The discoverer is responsible for global searching and prioritizes exploring unvisited areas, while the followers are used for local development near the discoverer.

4. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 3, characterized in that, In Step 2, the path point must meet the minimum safe distance constraint from the obstacle, that is, the distance from the current position of the drone to the obstacle must be greater than the preset minimum safe distance, and the total path length must not exceed the preset threshold.

5. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 4, characterized in that, The multimodal information includes visual images, ultrasonic distance data, and IMU (Inertial Measurement Unit) data. The visual sensor captures image information in front of the UAV, the ultrasonic sensor measures the distance between the UAV and the nearest obstacle, and the IMU integrates accelerometers, gyroscopes, and magnetometers to measure the attitude and motion state of the UAV, thus obtaining accelerometer, gyroscope, and magnetometer data.

6. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 5, characterized in that, The sparrow search algorithm includes the following steps: Step 1: Initialize the sparrow population: First, randomly generate a set of candidate solutions around the drone, that is, the positions of the sparrows. Each candidate solution represents a possible flight path. Step 2: Calculate the fitness value: For each candidate solution, calculate its fitness value, which is a comprehensive evaluation of the distance between the UAV and obstacles and the path length when flying along the path. Step 3: Update sparrow positions: According to the rules of the sparrow search algorithm, update the sparrow positions. The discoverer is responsible for the global search, and the follower is responsible for the local search. Through continuous iteration, the sparrow population gradually approaches the global optimum. Step 4: Path optimization: Use the globally optimal solution found by the sparrow search algorithm as the flight path of the UAV.

7. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 6, characterized in that, The operation steps of the data processing module are as follows: S1. Visual Image Processing: First, visual image processing uses the YOLOv5 algorithm to identify obstacles and obtain their two-dimensional image coordinates. Combining the camera intrinsic parameter matrix and the distance value measured by the ultrasonic sensor, the image coordinates are converted into three-dimensional coordinates (x, y, z) in the UAV coordinate system, and the distance value is output. S2. Ultrasonic data processing: The output distance value is smoothed by moving average filtering technology, and the distance value measured by ultrasound is cross-validated by the results of visual image detection. When the distance deviation between visual and ultrasonic detection exceeds the preset threshold, data correction is triggered. S3, IMU data fusion: Complementary filtering is used to integrate accelerometer, gyroscope and magnetometer data to estimate the UAV's attitude angle, speed and altitude. At this time, the UAV motion state output by the IMU forms a spatial correlation with the obstacle distance data optimized in step S2, and the relative motion trend of the obstacle is predicted based on the UAV speed. S4. Status Assessment: By setting attitude angle thresholds and speed fluctuation ranges, the system determines whether the UAV is in a stable flight state. At the same time, the altitude data output by the IMU verifies the vertical accuracy of the ultrasonic ranging. The system integrates the obstacle's three-dimensional coordinates, distance, speed, and the UAV's own status parameters from steps S1-S3 into the input list required by the path planning and obstacle avoidance module.

8. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 7, characterized in that, The attitude angles include the UAV's roll, pitch, and yaw. The UAV's speed is determined by integrating accelerometer data, and the UAV's altitude data is determined by fusing barometer data.

9. The multimodal information collaborative UAV omnidirectional autonomous obstacle avoidance algorithm according to claim 8, characterized in that, The control execution module includes discretizing the path planning results into time series trajectory points, generating target attitude commands through a PID algorithm, calculating the differential speed of the motor and adding feedforward compensation, and monitoring IMU data in real time. When the UAV attitude angle deviation exceeds a preset threshold, it automatically triggers the damping coefficient adjustment to maintain flight stability.

Images