An unmanned aerial vehicle adaptive path planning method fusing environmental uncertainty

Abstract

The application provides an unmanned aerial vehicle adaptive path planning method and system fusing environmental uncertainty, acquires surrounding environment data based on an environment perception module, generates a local grid map and a dynamic obstacle list, inputs an environmental uncertainty evaluation module, obtains an uncertainty factor, performs path planning through a global and local path planner based on the uncertainty factor, adjusts the weight of the global path and the local path according to the environmental uncertainty factor through a dynamic weight fuser, and obtains a final trajectory. The method effectively improves the flexibility and real-time performance of the unmanned aerial vehicle path planning, enhances the obstacle avoidance capability and flight stability of the unmanned aerial vehicle in a complex environment, and improves the flight efficiency.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicles (UAVs), particularly to the field of UAV path planning, and more specifically to a method and system for adaptive path planning of UAVs that incorporates environmental uncertainties. Background Technology

[0002] With the continuous development of drone technology, its applications in various fields are gradually increasing, especially in logistics delivery, geographic surveying, and environmental monitoring. Drones can complete tasks in these complex and ever-changing environments. The widespread application of drones mainly relies on their powerful path planning capabilities. Existing technologies for drone path planning generally divide them into two types: global path planning and local path planning.

[0003] Global path planning refers to the path design performed by the UAV before flight. It is usually based on known map information and uses algorithms to determine the optimal path from the starting point to the destination. Factors considered in this process include the maximum flight distance, minimum flight time, and obstacle avoidance. Local path planning, on the other hand, addresses real-time environmental changes during flight, such as encountering sudden obstacles or weather changes. In these cases, the UAV needs to readjust its flight path based on real-time perception information.

[0004] In actual flight, these two path planning methods need to work together. Global path planning is responsible for guiding the macroscopic route, while local path planning continuously fine-tunes the flight path based on environmental changes. When there is a conflict between the control intentions of the global and local paths, it may affect the local obstacle avoidance effect, leading to obstacle avoidance failure or path oscillation. This can prevent the drone from accurately avoiding obstacles, or even cause collisions. It may also cause large fluctuations in the drone's speed, making it unable to accurately maintain within the target speed range. Especially in complex environments or high-speed flight, this instability may lead to serious consequences such as drone crashes. Therefore, it is necessary to optimize the path planning method to improve the drone's performance.

[0005] Existing technical solutions often employ two processing approaches: layered architecture and prioritization. Layered architecture focuses on task decomposition and hierarchical information flow. The global planner is responsible for planning a macroscopic path, while the local planner uses this path as a "guide," combining real-time sensor data for fine-tuning and obstacle avoidance, and generating the final instructions for the flight controller. The global path is generally not updated frequently unless the local planner finds the current global path segment completely infeasible, or the mission objective undergoes a significant change.

[0006] Existing technologies for path planning, including layered architecture schemes and priority schemes, have the following drawbacks: 1) Layered architectures typically pre-define the hierarchical structure of task decomposition and information flow during the design phase, making it difficult to adjust the system when faced with sudden tasks or changes in task requirements. 2) Information typically needs to pass through multiple levels from the bottom-level sensors to the high-level planner, and then from the high-level planner to the bottom-level executor. This multi-level information transmission may lead to delays, especially in scenarios with high real-time requirements.

[0007] In addition, the priority scheme also presents the following challenges: 1) Difficulty in defining priorities: Setting priorities reasonably is a challenge. Fixed priorities may not be suitable for all situations. For example, in some tasks, closely tracking the path may be more important than minor obstacle avoidance (such as accurate mapping). Dynamically adjusting the priority logic will increase the system complexity. 2) Control conflicts and oscillations: If the priority switching logic is poorly designed, or if the instructions between modules differ too much, the UAV may frequently switch between different control objectives, resulting in unsmooth or even dangerous jittering or oscillation behavior. For example, if the obstacle avoidance module tells the UAV to go left, and the path tracking module wants it to go right back to the path, it may cause back-and-forth swaying.

[0008] It is evident that existing UAV path planning systems primarily employ hierarchical architectures or priority control strategies, both of which have limitations in practical applications. Hierarchical architectures are rigid in task decomposition and information flow transmission, making it difficult to flexibly respond to sudden tasks or environmental changes, and the latency caused by multi-level information transmission affects the system's real-time performance. Priority-based schemes, on the other hand, face difficulties in priority setting, control conflicts, and oscillations, resulting in insufficient stability and safety of UAV path planning in complex or dynamic environments. Summary of the Invention

[0009] The present invention aims to overcome the shortcomings of the prior art and improve the path planning capability and flight safety of UAVs in complex and dynamic environments.

[0010] The first aspect of this invention discloses an adaptive path planning method for unmanned aerial vehicles (UAVs) that incorporates environmental uncertainties, the method comprising:

[0011] Step S1: Set up the environmental perception module to acquire surrounding environmental data in real time through different sensors, and generate a local grid map and a dynamic obstacle list;

[0012] Step S2: Input the generated local grid map and dynamic obstacle list into the environmental uncertainty assessment module. The local grid map stores static environmental information and is used to calculate the static obstacle density and the proportion of unknown areas. The dynamic obstacle list stores the real-time status of dynamic obstacles and is used to calculate the dynamic obstacle threat level. The environmental uncertainty assessment module calculates the environmental uncertainty factor based on the static obstacle density, dynamic obstacle threat level, and the proportion of unknown areas.

[0013] Step S3: Incorporate a heuristic function and an environmental uncertainty factor into the global path planning algorithm. The global path planner uses the global path planning algorithm to output a global trajectory based on the UAV's starting position and target position; and generates a local trajectory based on the dynamic obstacle list, the global trajectory output by the global path planner, and the UAV's state information.

[0014] Step S4: Adjust the weights of the global trajectory and the local trajectory according to the environmental uncertainty factor to generate the final flight trajectory.

[0015] According to the UAV adaptive path planning method for incorporating environmental uncertainties as described in the first aspect of the present invention, the generation of the local grid map specifically involves: acquiring an original point cloud using a lidar sensor; transforming the original point cloud using coordinates, segmenting the ground, discretizing it into grids, and classifying its state to output a local grid map; dividing the UAV's flight area into grid cells; recording the spatial state of each grid cell; and updating the local grid map using the following formula:

[0016]

[0017] Where (i,j) are the coordinates of the grid cell, 1 indicates that the grid cell is an obstacle, 0 indicates that the grid cell is a passable area, and -1 indicates that the grid cell is an unknown area.

[0018] According to the UAV adaptive path planning method for incorporating environmental uncertainties as described in the first aspect of the present invention, generating a dynamic obstacle list specifically involves: relying on a binocular camera and LiDAR, and using data fusion and target detection algorithms to track the state of dynamic obstacles in real time; the state of the d-th dynamic obstacle is described by the following formula:

[0019] DOL(d)={(x d (t),y d (t)),V d (t),r d}

[0020] Where: (x d (t),y d (t) represents the position of the obstacle at time t, V d(t) is the velocity of the obstacle, r d It refers to the size of the obstacle.

[0021] According to the UAV adaptive path planning method integrating environmental uncertainties as described in the first aspect of the present invention, the calculation of the environmental uncertainty factor based on static obstacle density, dynamic obstacle threat level, and the influence of unknown areas in step S2 specifically involves:

[0022] (1) Static obstacle density

[0023] To quantify the distribution density of static obstacles in a local raster map, let the total number of raster cells be M×N, where M represents the number of raster cells in each row and N represents the number of raster cells in each column. The number of obstacle cells in the generated raster map is counted as C. obs Then the static obstacle density D obs for:

[0024]

[0025] The contribution of dynamic obstacles to the uncertainty factor needs to take into account the obstacle's velocity V. d (t), size r d And the relative distance to the drone;

[0026]

[0027] Where: K is the total number of dynamic obstacles, d u (t) represents the minimum distance between the dynamic obstacle and the drone, calculated using Euclidean distance;

[0028] (3) Proportion of unknown areas

[0029] Unknown region ratio U area Calculated by the proportion of unknown area grid cells in a local raster map:

[0030]

[0031] C unk This represents the number of cells in an unknown area of ​​a raster map. It iterates through each cell in the raster map. U area The proportions reflect the unperceived or uncertain aspects of the flight environment;

[0032] (4) Environmental uncertainty factors

[0033] The environmental uncertainty factor (EUM) is calculated by comprehensively considering the static obstacle density, dynamic obstacle threat level, and proportion of unknown areas in the environment.

[0034] EUM = w1·D obs +w2·Ddyn +w3·U area

[0035] Among them, D obs It is the static obstacle density in the environment, that is, the number of obstacles per unit area, calculated from the local grid map, D. dyn This refers to the dynamic obstacle threat level, which is the number of dynamic obstacles per unit area. It is calculated from a local grid map of the dynamic obstacle list. area is the proportion of unknown areas in the environment, representing areas that cannot be obtained by sensors. w1, w2, and w3 are weighting coefficients that must satisfy w1 + w2 + w3 = 1, representing the degree of contribution of each factor to the overall uncertainty assessment.

[0036] According to the first aspect of the present invention, the adaptive path planning method for UAVs that incorporates environmental uncertainties includes the following specific functions in step S3: cost function design, node expansion, and global path generation.

[0037] According to the first aspect of the present invention, the adaptive path planning method for unmanned aerial vehicles that incorporates environmental uncertainties includes, in step S3, the specific functions of the local path planner include obstacle detection and avoidance, dynamic path adjustment, and local trajectory generation.

[0038] According to the UAV adaptive path planning method for incorporating environmental uncertainties as described in the first aspect of the present invention, step S4 specifically includes:

[0039] The weight calculation formula is expressed as follows:

[0040]

[0041] w l =1-w g

[0042] Let the weight of the global trajectory be w. g The weight of the local trajectory is w l The parameter α is adjusted to regulate the system's sensitivity to environmental changes.

[0043] For global path point P g and local path point P l Weighted fusion is performed to form the final flight trajectory P. final :

[0044] P final =w g ·P g +w l ·P l .

[0045] A second aspect of the present invention discloses an adaptive path planning system for unmanned aerial vehicles (UAVs) that incorporates environmental uncertainties. The system includes a processing unit configured to perform steps for implementing the adaptive path planning method for UAVs that incorporates environmental uncertainties as described in the first aspect.

[0046] The third aspect of the present invention discloses an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the program in the memory to implement the UAV adaptive path planning method for incorporating environmental uncertainties as described in the first aspect.

[0047] The fourth aspect of the present invention discloses a computer-readable storage medium that stores the implementation of the UAV adaptive path planning method for incorporating environmental uncertainties as described in the first aspect.

[0048] In summary, this invention proposes an adaptive path planning method for UAVs that incorporates environmental uncertainties, which brings the following benefits:

[0049] (1) Achieve dynamic coordination between global and local paths, ensuring the overall mission objectives of the UAV while flexibly responding to sudden obstacles and environmental changes during flight, and avoiding obstacle avoidance failure or path oscillation caused by conflicting control intentions.

[0050] (2) The path planning strategy is dynamically adjusted according to the complexity of the environment, so that the system can prioritize the global path when the environment is simple to improve flight efficiency; when the environment is complex or uncertain, the weight of the local path is increased to improve obstacle avoidance capability and flight safety.

[0051] (3) Quantify environmental uncertainty and introduce it into path planning decision-making. By designing an environmental uncertainty factor (EUM), the static obstacle density, dynamic obstacle threat level and unknown area ratio are comprehensively considered to achieve real-time assessment of environmental complexity, and the fusion weight of global and local paths is dynamically adjusted accordingly.

[0052] (4) Improve the real-time performance and stability of the path planning system, reduce the delay caused by multi-level information transmission, avoid control conflicts and flight jitter caused by improper priority switching, and ensure that the UAV can complete the mission smoothly and efficiently in complex environments.

[0053] By solving the above-mentioned technical problems, the present invention can significantly improve the flexibility, real-time performance and safety of UAV path planning in complex and dynamic environments, overcome the limitations of existing technologies, and meet the needs of UAVs for efficient obstacle avoidance and stable flight in practical applications. Attached Figure Description

[0054] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0055] Figure 1 This is an execution flowchart of existing technologies based on priority schemes;

[0056] Figure 2 This is an execution flowchart of existing technologies based on a layered architecture scheme;

[0057] Figure 3 This is a flowchart illustrating the UAV adaptive path planning method for incorporating environmental uncertainties according to the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0059] The following is an explanation of some of the technical terms used in this article:

[0060] LGM (Local Grid Map) is a two-dimensional / three-dimensional environmental map built in real time by sensors (such as LiDAR). It divides the space into grid cells and marks them as obstacles, passable or unknown areas.

[0061] DOL (Dynamic Obstacle List) records real-time information about dynamic obstacles in the flight environment.

[0062] EUM (Environmental Uncertainty Metric) is an indicator that quantifies environmental complexity. It integrates static obstacle density, dynamic obstacle threat level, and the proportion of unknown areas to dynamically adjust path planning strategies.

[0063] Dynamic window method: A local path planning algorithm that searches for the optimal obstacle avoidance trajectory in the velocity space based on UAV dynamic constraints and real-time sensor data.

[0064] like Figure 1This diagram illustrates a priority-based approach in existing technology. This approach continuously monitors the environment and the drone's status. When a specific condition is triggered (e.g., obstacle detection), the higher-priority module (usually obstacle avoidance) takes control. When the condition is resolved, control may revert to a lower-priority module (e.g., path tracking). Priority-based approaches emphasize immediate decision-making and conflict resolution, dynamically switching control modules based on priority by monitoring the environment and drone status, making them suitable for dynamic environments.

[0065] like Figure 2 This diagram illustrates a layered architecture approach in existing technologies. After starting, the process begins with a global planning layer to generate a global path, then proceeds to a local planning layer for process flight control execution. As can be seen, this method, through the collaborative work of the global and local planners, decomposes the task globally and passes information layer by layer, making it suitable for scenarios with relatively stable environments and accurate prior information.

[0066] like Figure 3 This is a schematic diagram of the UAV adaptive path planning method based on environmental uncertainty proposed in this invention.

[0067] Step S1: Configure the environmental perception module

[0068] First, the drone acquires data about its surrounding environment using various sensors, including LiDAR and binocular cameras. These sensors collect environmental information in real time, generating a Local Grid Map (LGM) and a Dynamic Obstacle List (DOL). The LGM includes occupancy status and space information, while the DOL includes information such as the location, speed, and size of obstacles.

[0069] The environmental perception module first collects environmental data through sensors (such as LiDAR and binocular cameras). The data from each sensor needs to be preprocessed and synchronized to ensure the accuracy and consistency of the data.

[0070] • LiDAR: Acquires distance information of surrounding obstacles and generates 3D point cloud data.

[0071] • Visual sensor: Uses a camera to acquire image data and combines it with deep learning algorithms to detect objects and identify dynamic obstacles.

[0072] • IMU: Provides acceleration and angular velocity data for the drone, and calculates the drone's motion state.

[0073] • GPS: Provides the drone's geographic location and altitude.

[0074] (1) Generation of local raster maps

[0075] A Local Grid Map (LGM) can be constructed using LiDAR, dividing the flight area into grid cells. Each cell records its spatial state (obstacles / accessibility / unknown). The LGM can be updated using the following formula:

[0076]

[0077] Where (i,j) are the coordinates of the grid. 1 represents an obstacle, 0 represents a passable area, and -1 represents an unknown area.

[0078] (2) Generation of dynamic obstacle list

[0079] Dynamic obstacle detection primarily relies on binocular cameras and LiDAR. Through data fusion and target detection algorithms, the state of dynamic obstacles can be tracked in real time. The state of the d-th dynamic obstacle can be described by the following formula:

[0080] DOL(d)={(x d (t),y d (t)),V d (t),r d}

[0081] Where: (x d (t),y d (t) represents the position of the obstacle at time t, V d (t) is the velocity of the obstacle, r d It refers to the size of the obstacle.

[0082] Step S2: Input the information acquired by the environmental perception module into the environmental uncertainty assessment module.

[0083] The Local Grid Map (LGM) stores static environmental information (such as obstacle locations, passable areas, and unknown areas), which is directly used to calculate the static obstacle density (Dobs) and the proportion of unknown areas (Uarea); the Dynamic Obstacle List (DOL) stores the real-time status of dynamic obstacles (position, speed, size, etc.), which is directly used to calculate the dynamic obstacle threat level (Ddyn).

[0084] The environmental information provided by the environmental perception module is input into the environmental uncertainty assessment module. The core task of this module is to quantify the complexity of the current environment and dynamically adjust the path planning strategy based on the environmental uncertainty. An environmental uncertainty factor (EUM) is designed based on static obstacle density, dynamic obstacle threat level, and the impact of unknown areas. This allows the system to assess the simplicity of the environment and adjust its dependence on global and local paths accordingly.

[0085] The Local Grid Map (LGM) and Dynamic Obstacle List (DOL) generated by the Environmental Perception Module are input into the Environmental Uncertainty Assessment Module. The assessment module relies on this raw data (LGM / DOL) to calculate subsequent quantitative indicators (Dobs, Ddyn, Uarea) and finally synthesize the EUM.

[0086] To quantify environmental uncertainty and dynamically adjust path planning strategies, an Environmental Uncertainty Factor (EUM) can be designed based on static obstacle density, dynamic obstacle threat level, and the impact of unknown areas. The EUM is a comprehensive quantitative indicator used to dynamically assess the complexity and risk level of the current flight environment.

[0087] (1) Static obstacle density

[0088] Quantify the distribution density of static obstacles in a local raster map. Let the total number of grid cells in the raster map be M×N, where M represents the number of grid cells in each row and N represents the number of grid cells in each column. Count the number of obstacle cells in the generated raster map as C. obs Then the static obstacle density D obs for:

[0089]

[0090] (2) Threat level of dynamic obstacles

[0091] The motion state of dynamic obstacles has a significant impact on the complexity of the environment. Therefore, the contribution of dynamic obstacles to uncertainty factors needs to take into account the obstacle's speed, size, and relative distance to the drone.

[0092]

[0093] Where: K is the total number of dynamic obstacles, d u (t) represents the minimum distance between the dynamic obstacle and the drone, calculated using Euclidean distance.

[0094] (3) The impact of unknown areas

[0095] The impact of unknown areas primarily stems from regions that cannot be fully perceived, potentially concealing obstacles or other dynamic changes. The proportion of unknown areas in Uarea can be calculated using the proportion of "unknown area" cells within the Local Grid Map (LGM).

[0096]

[0097] C unk This represents the number of cells in an unknown area of ​​a raster map. It iterates through each cell in the raster map. U areaThis ratio reflects the unperceived or uncertain parts of the flight environment; the larger this area, the higher the environmental uncertainty.

[0098] (4) Environmental uncertainty factors

[0099] It comprehensively considers factors such as obstacle density in the environment, the motion state of dynamic obstacles, and the proportion of unknown areas. Its calculation formula can be defined as follows:

[0100] Where: EUM = w1·D obs +w2·D dyn +w3·U area

[0101] ·D obs It is the obstacle density in the environment, that is, the number of obstacles in a unit area, which is usually calculated from a local grid map (LGM).

[0102] ·D dyn It is the dynamic obstacle threat level, which is the number of dynamic obstacles in a unit area, usually calculated from the Dynamic Obstacle List (DOL).

[0103] ·U area It represents the proportion of unknown areas in the environment, indicating areas that cannot be detected by sensors. It is usually calculated by the number of unknown areas in the LGM (Local Level Meter).

[0104] • w1, w2, w3 are weighting coefficients, which must satisfy w1 + w2 + w3 = 1, representing the degree of contribution of each factor to the overall uncertainty assessment.

[0105] Step S3: Based on the output of the environmental uncertainty assessment module, set up the global path planner and the local path planner.

[0106] Based on environmental perception and uncertainty assessment, passable areas are extracted using LGM (Locally Accessible Metrics), and a heuristic function is added to the global path planning algorithm to incorporate the influence of EUM (Earning Uncertainty). The global path planner outputs a global trajectory P based on the UAV's starting and target positions. g .

[0107] The task of the local path planner is to adjust and optimize the flight trajectory in real time to avoid dynamic obstacles encountered during flight. This module generates the local trajectory (PL) based on the Dynamic Obstacle List (DOL), the path trajectory output by the global path planner, and the UAV's state information.

[0108] The global path planner's specific functions include cost function design, node expansion, and global path generation. The global path planner generates global paths based on the improved A* algorithm or the RRT algorithm, and adjusts the conservatism of the paths by incorporating the environmental uncertainty factor (EUM). This invention uses the improved A* algorithm as an example for illustration:

[0109] Step S31: Extracting passable areas

[0110] Based on Local Grid Map (LGM), the environment is divided into passable areas (marked as 0), obstacle areas (marked as 1), and unknown areas (marked as -1). After initializing the starting point and target point and removing obstacle and unknown areas, a set of passable nodes is constructed.

[0111] Step S32: Cost Function Design

[0112] Heuristic weights are assigned to each candidate path node, and the environmental uncertainty factor (EUM) of the current node is incorporated into the heuristic function, adding the influence of EUM. The cost function is:

[0113] f(n)=g(n)+h(n)·(1+λ·EUM(n))

[0114] Where g(n): the actual cost (e.g., flight distance) from the starting point to node n.

[0115] h(n): Heuristic cost (Euclidean distance) from node n to the endpoint.

[0116] ·λ: Dynamic weighting coefficient.

[0117] •EUM(n): The EUM value of the region where node n is located.

[0118] Step S33: Node Expansion and Path Generation

[0119] 1) Starting from the starting point, add adjacent passable nodes to the open list and calculate f(n) for each node;

[0120] 2) Each time, select the node with the smallest f(n) in the open list to expand, and update the cost of its neighboring nodes;

[0121] 3) If the EUM value of the current node's grid exceeds the threshold (e.g., EUM(n) > 0.7), then increase the penalty weight for the path bypassing the area;

[0122] 4) Traverse all passable nodes on the map and generate a global path Pg.

[0123] The local path planner's specific functions include obstacle detection and avoidance, dynamic path adjustment and local trajectory generation. The local planner combines DOL data to generate emergency obstacle avoidance local trajectories. It can be designed using algorithms such as Dynamic Window DWA and artificial potential fields. This case study uses the artificial potential field method for illustration.

[0124] Step S34: Set up the gravitational potential field

[0125] The gravitational potential field of the target point on the drone is defined as follows:

[0126]

[0127] Where q is the current position of the drone, q d Let d(q,q) be the target point position. d ) represents the distance from the drone to the d-th obstacle; k att This is the gravitational coefficient, used to control the strength of convergence towards the target.

[0128] Step S35: Set up a repulsive potential field

[0129]

[0130] Where: k rep d is the repulsive force coefficient, θ is the angle between the relative motion direction of the UAV and the obstacle, and d is the repulsive force coefficient. safe =r d +r u +δ, EUM represents the safety distance (the meaning is inconsistent), r u Where is the radius of the UAV, δ is the adjustment coefficient, and η is the directional sensitivity coefficient;

[0131] Step S36: Calculation of resultant force and direction of motion

[0132] The net force acting on the drone is the vector sum of the gravitational force and all repulsive forces:

[0133]

[0134] Step S37: Update the motion model

[0135] Generate velocity command v based on the direction of the resultant force. cmd

[0136]

[0137] Where: the constraint condition is the maximum velocity amplitude v max The direction is the same as the resultant force, passing through the currently known position q. current The position q at the next moment is calculated by integration. next :

[0138] qnext =q current +v cmd ·Δt

[0139] Step S38: Generate local trajectory points

[0140] The drone's motion is calculated over multiple time steps to obtain a series of position and velocity updates, which are then combined to form a local trajectory.

[0141] Step S4: Dynamic Weight Fusion

[0142] When the environment becomes more complex, the system needs to adjust the global trajectory P according to the environmental uncertainty factor (EUM). g and local trajectory P l The weights are used to generate the final flight trajectory. The dynamic weight fusion unit adjusts the global trajectory P... g and local trajectory P l The weighting of the parameters ensures a balance between safety and efficiency for the drone, generates the final trajectory, and sends it to the flight control system for execution.

[0143] The weights of the global and local paths are dynamically adjusted based on the Environmental Uncertainty Factor (EUM). A lower EUM indicates a simpler environment and higher reliability of the global path; therefore, the weight of the global path can be increased while the weight of the local path decreases. Conversely, a higher EUM indicates a more complex environment, where real-time performance and flexibility are more important for the local path; therefore, the weight of the local path can be increased while the weight of the global path decreases. The weight adjustment formula can be expressed as:

[0144] Weight calculation

[0145]

[0146] w l =1-w g

[0147] Let the weight of the global trajectory be w. g The weight of the local trajectory is w l Adjusting parameter α can regulate the system's sensitivity to environmental changes.

[0148] (2) Trajectory fusion

[0149] For global path point P g and local path point P l Weighted fusion is performed to form the final trajectory P. final The merged flight trajectory Pfinal is sent to the flight control system at a certain frequency. The flight control system generates specific speed, attitude and other control commands based on the trajectory, driving the UAV to fly safely and efficiently along the planned path.

[0150] P final =w g ·P g +w l ·P l

[0151] The present invention also provides a fine-grained radar operating mode recognition system based on semi-supervised contrastive learning, which is used to implement the aforementioned fine-grained radar operating mode recognition method based on semi-supervised contrastive learning.

[0152] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the program in the memory to implement the aforementioned fine-grained radar operating mode recognition method based on semi-supervised contrastive learning.

[0153] The present invention also provides a computer-readable storage medium storing a computer program that implements the aforementioned fine-grained radar operating mode recognition method based on semi-supervised contrastive learning.

[0154] In summary, this invention effectively improves the flexibility and real-time performance of UAV path planning by adjusting the weights of global and local paths based on environmental uncertainty factors through a dynamic weight fusion processor. This enhances the UAV's obstacle avoidance capabilities and flight stability in complex environments, thereby improving flight efficiency.

[0155] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for adaptive path planning of unmanned aerial vehicles (UAVs) that incorporates environmental uncertainties, characterized in that, The method includes: Step S1: Set up the environmental perception module to acquire surrounding environmental data in real time through different sensors, and generate a local grid map and a dynamic obstacle list; Step S2: Input the generated local grid map and dynamic obstacle list into the environmental uncertainty assessment module. The local grid map stores static environmental information and is used to calculate the static obstacle density and the proportion of unknown areas. The dynamic obstacle list stores the real-time status of dynamic obstacles and is used to calculate the dynamic obstacle threat level. The environmental uncertainty assessment module calculates the environmental uncertainty factor based on the static obstacle density, dynamic obstacle threat level, and the proportion of unknown areas. Step S3: Incorporate a heuristic function and an environmental uncertainty factor into the global path planning algorithm. The global path planner uses the global path planning algorithm to output a global trajectory based on the UAV's starting position and target position. The local path planner generates a local trajectory based on the dynamic obstacle list, the global trajectory output by the global path planner, and the UAV's state information. Step S4: Adjust the weights of the global trajectory and the local trajectory according to the environmental uncertainty factor to generate the final flight trajectory.

2. The UAV adaptive path planning method for incorporating environmental uncertainties as described in claim 1, characterized in that, The specific steps for generating the local grid map are as follows: First, acquire the original point cloud using LiDAR. Then, transform the original point cloud using coordinates, segment the ground, discretize it into grids, and classify its state to output a local grid map. Divide the UAV's flight area into grid cells, with each grid cell recording its spatial state. Update the local grid map using the following formula: Where (i,j) are the coordinates of the grid cell, 1 indicates that the grid cell is an obstacle, 0 indicates that the grid cell is a passable area, and -1 indicates that the grid cell is an unknown area.

3. The UAV adaptive path planning method for incorporating environmental uncertainties as described in claim 2, characterized in that, The generation of the dynamic obstacle list specifically involves: relying on a binocular camera and LiDAR, and using data fusion and target detection algorithms to track the state of dynamic obstacles in real time; the state of the d-th dynamic obstacle is described by the following formula: in: It is an obstacle in time t The position at that time V d ( t ) is the speed of the obstacle. It refers to the size of the obstacle.

4. The UAV adaptive path planning method for incorporating environmental uncertainties according to claim 3, characterized in that, The calculation of the environmental uncertainty factor in step S2 based on static obstacle density, dynamic obstacle threat level, and the impact of unknown areas is specifically as follows: (1) Static obstacle density To quantify the distribution density of static obstacles in a local raster map, let the total number of raster cells be M×N, where M represents the number of raster cells in each row and N represents the number of raster cells in each column. Count the number of obstacle cells in the generated raster map as follows: C obs Then the static obstacle density D obs for: ； (2) Threat level of dynamic obstacles The contribution of dynamic obstacles to uncertainty factors needs to take into account the speed of the obstacles. V d ( t ),size r d And the relative distance to the drone; Where: K is the total number of dynamic obstacles. d u (t) This represents the minimum distance between a dynamic obstacle and the drone, calculated using Euclidean distance. (3) Proportion of unknown areas Unknown region ratio U area Calculated by the proportion of unknown area grid cells in a local raster map: C unk This represents the number of cells in an unknown area of ​​the raster map. It iterates through each cell in the raster map. U area The proportions reflect the unperceived or uncertain aspects of the flight environment; (4) Environmental uncertainty factors The environmental uncertainty factor is calculated by comprehensively considering the static obstacle density, dynamic obstacle threat level, and proportion of unknown areas in the environment. EUM : in, D obs This refers to the static obstacle density in the environment, i.e., the number of obstacles per unit area, calculated from a local raster map. D dyn This refers to the dynamic obstacle threat level, which is the number of dynamic obstacles per unit area. It is calculated from a local grid map of the dynamic obstacle list. U area is the proportion of unknown areas in the environment, representing areas that cannot be obtained by sensors. w1, w2, and w3 are weighting coefficients that must satisfy w1+w2+w3=1, representing the degree of contribution of each factor to the overall uncertainty assessment.

5. The UAV adaptive path planning method for incorporating environmental uncertainties according to claim 4, characterized in that, The specific functions of the global path planner in step S3 include cost function design, node expansion, and global path generation.

6. The UAV adaptive path planning method for incorporating environmental uncertainties according to claim 5, characterized in that, The specific functions of the local path planner in step S3 include obstacle detection and avoidance, dynamic path adjustment, and local trajectory generation.

7. The UAV adaptive path planning method for incorporating environmental uncertainties according to claim 6, characterized in that, Step S4 specifically includes: The weight calculation formula is expressed as follows: Let the weight of the global trajectory be... The weights of the local trajectories are Adjust parameters It is used to adjust the system's sensitivity to environmental changes; global path points and local path points Weighted fusion is performed to form the final flight trajectory. : 。 8. An adaptive path planning system for unmanned aerial vehicles (UAVs) that integrates environmental uncertainties, characterized in that, The system includes a processing unit configured to perform the steps of implementing the UAV adaptive path planning method for incorporating environmental uncertainties as described in any one of claims 1-7.

9. An electronic device, characterized in that: The system includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the program in the memory to implement the UAV adaptive path planning method for incorporating environmental uncertainties as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores the UAV adaptive path planning method for incorporating environmental uncertainties as described in any one of claims 1-7.

Images