Intelligent logistics distribution path planning system for drones

Abstract

The present invention relates to the field of unmanned aerial vehicle logistics distribution technology, and specifically to an intelligent logistics distribution path planning system for unmanned aerial vehicles, including a three-dimensional geographic information acquisition module, a dynamic obstacle detection module, a hierarchical path planning module, a real-time environment perception module, and a path adjustment module; wherein: the three-dimensional geographic information acquisition module: is used to collect three-dimensional geographic information in the distribution area; the dynamic obstacle detection module: generates the location information and motion trajectory data of obstacles; the hierarchical path planning module: generates a multi-level flight path; the real-time environment perception module identifies the potential risks that affect the flight path in the current environment; the path adjustment module: adjusts the flight path of the unmanned aerial vehicle in real time. The present invention optimizes the flight path through real-time environment perception, dynamic obstacle detection, and autonomous learning algorithms, ensuring that the unmanned aerial vehicle can flexibly avoid obstacles and adapt to meteorological changes in a complex urban environment, thereby achieving efficient and safe logistics distribution.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle logistics distribution, and in particular to an intelligent logistics distribution path planning system for unmanned aerial vehicles. Background Art

[0002] With the rapid development of drone technology, the application of intelligent drones in the field of logistics and distribution has gradually increased, especially in complex urban environments, where drone delivery has the advantages of high efficiency and speed. However, the complex urban environment brings many challenges, such as dense high-rise buildings, flight restricted areas, dynamic obstacles and other factors, which bring great difficulties to the path planning and flight safety of drones. In order to ensure that drones can efficiently complete delivery tasks at different altitudes, traditional path planning algorithms are often limited to static environments and lack the ability to respond to real-time environmental changes and dynamic risks. In addition, the meteorological conditions in the environment, such as wind speed, wind direction, and air pressure changes, also have a very obvious impact on drone flight. How to comprehensively consider three-dimensional geographic information, obstacle distribution and meteorological conditions in complex urban environments to generate safe and dynamically adjusted multi-level flight paths has become a technical problem that needs to be solved urgently in intelligent logistics and distribution.

[0003] Existing path planning systems are usually based only on preset geographic information and flight rules, lacking the ability to perceive and dynamically adjust real-time environmental changes, especially when faced with sudden obstacles and adverse weather conditions, making it difficult to flexibly avoid them, resulting in reduced flight path safety and efficiency. When UAVs fly in complex urban environments, they need to accurately perceive and integrate three-dimensional geographic information, combine dynamic data of obstacles and real-time weather information, and intelligently plan and adjust paths to cope with complex flight scenarios. Therefore, it is of great technical necessity to propose an intelligent logistics distribution path planning system for UAVs to improve the safety and efficiency of UAVs in urban logistics distribution. Summary of the invention

[0004] Based on the above objectives, the present invention provides an intelligent logistics distribution path planning system for unmanned aerial vehicles.

[0005] The intelligent logistics distribution path planning system of UAV includes a three-dimensional geographic information acquisition module, a dynamic obstacle detection module, a hierarchical path planning module, a real-time environment perception module and a path adjustment module; among which:

[0006] 3D geographic information collection module: Deploy lidar and cameras on drones to collect 3D geographic information on building height, location, shape and distribution within the delivery area;

[0007] Dynamic obstacle detection module: Combines the 3D geographic information provided by the 3D geographic information acquisition module and the dynamic information collected by the drone sensor to detect moving obstacles on the delivery path in real time, and generates the location information and motion trajectory data of the obstacles, which are then transmitted to the hierarchical path planning module;

[0008] Hierarchical path planning module: Based on the 3D geographic information provided by the 3D geographic information acquisition module and the obstacle information transmitted by the dynamic obstacle detection module, combined with factors such as flight altitude, building distribution and no-fly zones, a multi-level flight path is generated;

[0009] Real-time environmental perception module: Through the meteorological sensors deployed on the drone, real-time meteorological data, including wind speed, wind direction and air pressure, are collected, and the real-time meteorological condition data is compared and analyzed with the generated multi-level flight path information to identify potential risks that affect the flight path in the current environment;

[0010] Path adjustment module: Based on the potential risks identified by the real-time environmental perception module, and combined with historical flight data and current environmental changes, the autonomous learning algorithm is used to adjust the flight path of the drone in real time to optimize the flight path.

[0011] Optionally, the three-dimensional geographic information acquisition module includes a laser radar unit, a camera unit and a data processing unit; wherein:

[0012] LiDAR unit: installed at the bottom of the drone, used to emit laser beams and receive signals reflected from the surface of buildings, and calculate the height and spatial position of buildings by measuring the signal return time;

[0013] Camera unit: installed on the front of the drone, used to capture the external contour and shape information of the building and generate a two-dimensional image of the building;

[0014] Data processing unit: electrically connected to the lidar unit and the camera unit, used to integrate the height and spatial position information generated by the lidar with the image data collected by the camera unit, and generate a three-dimensional geographic model of the delivery area.

[0015] Optionally, the data processing unit specifically includes:

[0016] Integrate height and spatial position information with 2D image data: Combine the building height and spatial coordinates generated by the lidar with the 2D image coordinates generated by the camera unit Match the pixels in each two-dimensional image Corresponding to a point in three-dimensional space , giving each pixel its real three-dimensional coordinates;

[0017] Generate 3D point cloud data: Generate the 3D coordinate point corresponding to each pixel through the matched height and 2D image data , and the three-dimensional coordinate points form a three-dimensional point cloud data set of the building, each point represents a position on the surface of the building;

[0018] Point cloud data optimization: based on 2D images taken from multiple angles, point cloud data can be stitched and redundant points can be eliminated;

[0019] Constructing a 3D model of a building: Generate the 3D geometric shape of the building by converting the optimized point cloud data into a triangular mesh model. Specifically, the volume and appearance of the building are represented by connecting adjacent point cloud data points into a mesh.

[0020] Generate a complete 3D geographic model of the delivery area: Spatially fuse the 3D models of each building to generate a complete 3D geographic model of the delivery area.

[0021] Optionally, the dynamic obstacle detection module includes an environment perception unit, an obstacle recognition unit and a motion trajectory prediction unit; wherein:

[0022] Environmental perception unit: used to receive the three-dimensional geographic information of the delivery area from the three-dimensional geographic information acquisition module, and combine with the camera and infrared sensor deployed on the drone to capture the real-time image and thermal signal of dynamic objects on the delivery path;

[0023] Obstacle recognition unit: It identifies moving obstacles in the delivery path by analyzing the images and infrared data collected by the environmental perception unit. It uses computer vision algorithms to process the images collected by the camera, extracts the edge, shape and position features of the obstacle, and verifies whether the obstacle exists by combining the thermal signal provided by the infrared sensor.

[0024] Motion trajectory prediction unit: Based on the obstacle position information provided by the obstacle recognition unit and combined with continuous image frames, the motion trajectory of the obstacle is calculated. The formula is:

[0025] ;

[0026] in, is the speed of the obstacle; Indicates the obstacle at time The three-dimensional space coordinates of the time, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; Indicates the obstacle at time The three-dimensional space coordinates of the time, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; They are all time points, used to describe the location information of obstacles at different times. .

[0027] Optionally, the obstacle recognition unit specifically includes:

[0028] Image preprocessing: grayscale the original image captured by the camera to reduce the color information of the image;

[0029] Edge detection: Use the Canny edge detection algorithm to extract the edges of the preprocessed grayscale image to extract the outline of the obstacle;

[0030] Shape feature extraction: Through the contour detection algorithm, the shape contour of the obstacle is extracted from the edge detection result, and its contour is represented as a series of point sets including ,The point set is used to form the two-dimensional boundary shape of the obstacle;

[0031] Position feature calculation: Based on the camera's intrinsic parameter matrix K and the extracted contour point set, the edge contour in the two-dimensional image is projected back to the three-dimensional space using inverse projection to calculate the position coordinates of the obstacle in the two-dimensional space. ;

[0032] Thermal signal verification: Combined with the thermal signal data collected by the infrared sensor, by setting the thermal signal threshold , to determine whether there is an obstacle; if the thermal signal value of a certain pixel satisfy: , it means there is an obstacle.

[0033] Optionally, the hierarchical path planning module includes a data receiving unit, a height level planning unit, an obstacle avoidance unit, a no-fly zone processing unit and a path generation unit; wherein:

[0034] Data receiving unit: used to receive the building height, position, shape and distribution data from the three-dimensional geographic information acquisition module, as well as the obstacle position information and motion trajectory data transmitted by the dynamic obstacle detection module. The three-dimensional geographic information is used to construct the basic spatial framework of the distribution area, and the obstacle information is used to assist the dynamic obstacle avoidance of the path;

[0035] Altitude planning unit: Based on the three-dimensional geographic information provided by the data receiving unit, the delivery area is divided into multiple flight altitude levels. Each altitude level is divided according to the height distribution of buildings to ensure that the drone can flexibly switch between different altitude levels to avoid collisions with buildings.

[0036] Obstacle avoidance unit: dynamically adjusts the flight path based on obstacle information, generates an adaptive obstacle avoidance path by analyzing the location information and movement trajectory of the obstacle; the calculation of the obstacle avoidance path is based on the shortest path algorithm to ensure that the drone can bypass the obstacle;

[0037] No-fly zone processing unit: According to the marked no-fly zone data in the area, the no-fly zone is marked as a no-entry area, and the drone is prevented from entering the no-fly zone during path planning. The constraints of the no-fly zone are set as follows: ,in, is the current coordinate of the drone, is the spatial coordinate of the no-fly zone;

[0038] Path generation unit: used to generate a multi-level flight path by combining the flight data generated by the altitude level planning unit, obstacle avoidance unit and no-fly zone processing unit.

[0039] Optionally, the path generation unit specifically includes:

[0040] Integrate flight data: receive altitude data provided by the altitude planning unit, obstacle avoidance path data provided by the obstacle avoidance unit, and no-fly zone coordinate data provided by the no-fly zone processing unit;

[0041] Path optimization: Based on the flight data provided by each unit, the path generation unit uses an optimization algorithm to generate the optimal path. By considering the flight altitude, obstacle location and no-fly zone boundaries, the shortest path algorithm is used to calculate the optimized path. ;

[0042] Flight altitude adjustment: Based on the flight altitude provided by the altitude planning unit, the path generation unit dynamically adjusts the altitude level of the flight path according to the distribution and height of the buildings.

[0043] Obstacle avoidance path generation: Combines the position and motion trajectory of the obstacle and generates a dynamic obstacle avoidance path by calculating the position change of the obstacle in the path ;

[0044] No-fly zone processing: The path generation unit adjusts the flight path according to the coordinate data of the no-fly zone to ensure that the UAV flight path is away from the no-fly zone;

[0045] Final path generation: Integrate the adjusted obstacle avoidance path, flight altitude level, and path data after no-fly zone processing to generate the final multi-level flight path , the formula is: ,in, is the final generated path, is the adjusted obstacle avoidance path, For the flight altitude.

[0046] Optionally, the real-time environment perception module includes a meteorological data acquisition unit, a data processing unit and a risk assessment unit; wherein:

[0047] Meteorological data collection unit: A meteorological sensor installed on the drone that collects real-time data on meteorological conditions in the environment during flight, including wind speed ,wind direction and air pressure , meteorological condition data are continuously collected by meteorological sensors and transmitted to the data processing unit;

[0048] Data processing unit: Receives real-time meteorological data provided by the meteorological data acquisition unit, and compares and analyzes it with the path point information in the multi-level flight path to determine whether the meteorological conditions have an impact on each point in the path; the specific analysis formula is:

[0049] ,in, is the distance between the flight path point and the center of wind direction change, are the center coordinates of the area affected by wind speed and direction, Represents the drone in the path Points Axis coordinates; Represents the drone in the path The Y-axis coordinate of a point; Represents the drone in the path Points Axis coordinates, in meters;

[0050] Risk Assessment Unit: By comparing and analyzing the results, combined with the current wind speed and wind direction data, the potential risks on the flight path are assessed; if it is detected that any point on the path is less than the safety threshold from the center of the wind speed change area , then mark the point as a risk point.

[0051] Optionally, the path adjustment module includes a risk input unit, a historical data analysis unit, an autonomous learning unit and a path output unit; wherein:

[0052] Risk input unit: receives risk data from the real-time environment perception module, including risk markers on the path points and the distance to the center of the wind speed variation area , and pass the path points marked as risks to the autonomous learning unit as trigger signals for path adjustment;

[0053] Historical data analysis unit: collects historical flight data, including weather conditions and path adjustment records encountered in previous flights, and predicts the changing pattern of the optimal flight path of the drone under similar weather conditions based on historical data;

[0054] Autonomous learning unit: uses autonomous learning algorithms to integrate the current environmental data provided by the real-time environmental perception module and the historical flight data provided by the historical data analysis unit, and adjusts the flight path through multiple iterations;

[0055] Path output unit: outputs the adjusted new path to the UAV's flight control system for updating flight instructions.

[0056] Optionally, the autonomous learning unit specifically includes:

[0057] Data input: Receive current environmental data from the real-time environmental perception module, including wind speed ,wind direction , air pressure and historical flight path adjustment data provided by the historical data analysis unit ;

[0058] Learning algorithm processing: Based on the received current environment data and historical data, the reinforcement learning algorithm is used to optimize the flight path of the drone. The reinforcement learning algorithm uses the safety and path optimization of the drone as the reward function, and continuously updates the optimal path strategy through data feedback from multiple flight missions. The optimization objective function of reinforcement learning is: ,in, is the cumulative reward value of the future path, is the immediate reward for each time step, is the discount factor, is the number of future path points;

[0059] Strategy update: During the learning process, the path adjustment strategy is continuously updated. Based on the feedback from reinforcement learning, the strategy update unit generates a new path adjustment direction. , the formula for path adjustment is: ,in, For the new path, is the current path, is the risk assessment function, and is the path adjustment factor.

[0060] Beneficial effects of the present invention:

[0061] The present invention, by integrating three-dimensional geographic information collection, dynamic obstacle detection, real-time environmental perception and autonomous learning algorithms, can generate and dynamically adjust multi-level flight paths suitable for drones in complex urban environments. The system uses lidar, cameras and other sensors to accurately obtain building information in the delivery area, combined with the real-time position and motion trajectory data of obstacles, to ensure that the drone can safely avoid dynamic obstacles. In addition, the real-time environmental perception module collects data such as wind speed, wind direction and air pressure through meteorological sensors. The system compares and analyzes this information with the flight path, identifies potential risks and adjusts the path in time, thereby effectively improving the safety of delivery tasks and the accuracy of path planning.

[0062] The present invention, through the introduction of autonomous learning units, enables the system to intelligently adjust the flight path according to historical flight data and current environmental changes, and continuously optimizes the path strategy through reinforcement learning algorithms to ensure that the UAV maintains flight efficiency and avoids risks in complex environments. On the basis of real-time path planning and environmental perception, the flight path is dynamically optimized to improve the adaptability of the UAV in logistics distribution tasks, solve the technical problems of the traditional system's lack of flexibility and dynamic response capabilities, and significantly improve the efficiency and safety of UAV distribution in urban environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 A schematic diagram of an intelligent logistics distribution path planning system according to an embodiment of the present invention;

[0065] Figure 2 Schematic diagram of a hierarchical path planning module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0067] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0068] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0069] like Figure 1-Figure 2 As shown, the intelligent logistics distribution path planning system of UAV includes a three-dimensional geographic information acquisition module, a dynamic obstacle detection module, a hierarchical path planning module, a real-time environment perception module and a path adjustment module; wherein:

[0070] 3D geographic information collection module: Deploy lidar and cameras on drones to collect 3D geographic information on building height, location, shape and distribution in the delivery area, providing basic geographic data for route planning;

[0071] Dynamic obstacle detection module: Combines the 3D geographic information provided by the 3D geographic information acquisition module and the dynamic information collected by the drone sensors (such as cameras and infrared sensors), detects moving obstacles on the delivery path in real time, generates the location information and motion trajectory data of the obstacles, and transmits them to the hierarchical path planning module;

[0072] Hierarchical Path Planning Module: Based on the 3D geographic information provided by the 3D Geographic Information Acquisition Module and the obstacle information transmitted by the Dynamic Obstacle Detection Module, and combined with factors such as flight altitude, building distribution and no-fly zones, it generates a multi-level flight path suitable for complex urban environments, and transmits the path information to the real-time environment perception module and the path adjustment module;

[0073] Real-time environmental perception module: Through the meteorological sensors deployed on the drone, real-time meteorological data, including wind speed, wind direction and air pressure, are collected, and the real-time meteorological condition data is compared and analyzed with the generated multi-level flight path information to identify potential risks that affect the flight path in the current environment;

[0074] Path adjustment module: Based on the potential risks identified by the real-time environmental perception module, and combined with historical flight data and current environmental changes, the autonomous learning algorithm is used to adjust the flight path of the drone in real time to optimize the flight path and ensure that the drone can complete the delivery task efficiently and safely in a complex urban environment.

[0075] The three-dimensional geographic information acquisition module includes a laser radar unit, a camera unit and a data processing unit; wherein:

[0076] LiDAR unit: installed at the bottom of the drone, used to emit laser beams and receive signals reflected from the surface of buildings. The height and spatial position of the building are calculated by measuring the signal return time. Specifically, the time it takes for the laser beam to be emitted to the surface of the building is , the return time is ,in is the speed of light in meters per second, is the distance between the drone and the building, in meters; the calculation formula is: , by measuring the distance and the current location coordinates of the drone , the height and spatial position of the building can be calculated;

[0077] Camera unit: Installed on the front of the drone, it is used to capture the external contour and shape information of the building and generate a two-dimensional image of the building; the camera captures images of the building from multiple angles through an image sensor and converts the three-dimensional spatial coordinates of the building into a real-time image through a pinhole camera model. Projected into 2D coordinates in the image plane , the process is described by the following formula:

[0078] ,

[0079] in, and is the two-dimensional pixel coordinate in the image plane, in pixels; is the coordinate of the building in three-dimensional space, in meters; is the scale factor, representing the projection ratio; the matrix is the intrinsic parameter matrix of the camera, including focal length and optical axis offset parameters; is the rotation matrix, is the translation vector, and the two define the posture information of the camera. Through this formula, the three-dimensional information of the building is projected into a two-dimensional image;

[0080] Data processing unit: electrically connected to the lidar unit and the camera unit, used to integrate the height and spatial position information generated by the lidar with the image data collected by the camera unit, and generate a three-dimensional geographic model of the delivery area, and transmit the three-dimensional geographic model to the dynamic obstacle detection module and the hierarchical path planning module; through the clear deployment of the lidar and camera units and their mutual cooperation, it can efficiently and accurately collect the three-dimensional geographic information of the buildings in the delivery area, provide precise height, position and shape data, and provide comprehensive and reliable basic information for subsequent path planning.

[0081] The three-dimensional geographic model generated in the distribution area in the data processing unit specifically includes:

[0082] Integrate height and spatial position information with 2D image data: Combine the building height and spatial coordinates generated by the lidar with the 2D image coordinates generated by the camera unit Match the pixels in each two-dimensional image Corresponding to a point in three-dimensional space , giving each pixel its real three-dimensional coordinates;

[0083] Generate 3D point cloud data: Generate the 3D coordinate point corresponding to each pixel through the matched height and 2D image data , and the three-dimensional coordinate points form a three-dimensional point cloud data set of the building, each point represents a position on the surface of the building. The process is through the formula: ,in is a point in a 3D point cloud;

[0084] Point cloud data optimization: Based on the two-dimensional images taken from multiple angles, the point cloud data is stitched and redundant points are eliminated. Specifically, the iterative closest point (ICP) algorithm is used to align the point cloud data from different angles to ensure the consistency and accuracy of the data. After the point cloud is optimized, the shape and surface features of the building are accurately presented; the goal of the ICP algorithm is to minimize the error between point clouds, align them in space, and optimize the accuracy of the point cloud. The formula of the ICP algorithm is: ,in, is the error function, is a point in the source point cloud, is the closest point in the target point cloud, is the total number of points, and the error is minimized through iterative adjustment , realize the alignment of point cloud data;

[0085] Constructing a 3D model of a building: Generate the 3D geometric shape of the building by converting the optimized point cloud data into a triangular mesh model. Specifically, the volume and appearance of the building are represented by connecting adjacent point cloud data points into a mesh. Specifically, the triangular mesh model connects adjacent point cloud data points into triangular patches, each of which consists of three points. The formula for generating the triangular mesh is: ,in represents a triangular patch, is the three-dimensional coordinates of the triangle vertices;

[0086] Generate a complete 3D geographic model of the delivery area: The 3D models of each building are spatially fused to generate a complete 3D geographic model of the delivery area. The model includes the height, location, appearance and spatial distribution of each building, providing an accurate 3D environmental reference for the drone's path planning.

[0087] The dynamic obstacle detection module includes an environment perception unit, an obstacle recognition unit, and a motion trajectory prediction unit; wherein:

[0088] Environmental perception unit: used to receive the three-dimensional geographic information of the delivery area from the three-dimensional geographic information acquisition module, and combined with the camera and infrared sensor deployed on the drone, to capture the real-time image and thermal signal of dynamic objects on the delivery path. The camera is used to obtain the two-dimensional image information of obstacles; the infrared sensor is used to detect the heat source signal of obstacles to ensure that obstacles can be identified even in low-light environments;

[0089] Obstacle recognition unit: It identifies moving obstacles in the delivery path by analyzing the images and infrared data collected by the environmental perception unit. It uses computer vision algorithms to process the images collected by the camera, extracts the edge, shape and position features of the obstacle, and verifies whether the obstacle exists by combining the thermal signal provided by the infrared sensor.

[0090] Motion trajectory prediction unit: Based on the obstacle position information provided by the obstacle recognition unit and combined with continuous image frames, the motion trajectory of the obstacle is calculated. The formula is:

[0091] ;

[0092] in, is the speed of the obstacle, in meters per second, indicating the speed at which the obstacle moves on the path; Indicates the obstacle at time The three-dimensional space coordinates of the time, in meters, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; Indicates the obstacle at time The three-dimensional space coordinates of the time, in meters, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; They are all time points in seconds, used to describe the location information of obstacles at different times. Through the collaborative work of the environmental perception unit, obstacle recognition unit and motion trajectory prediction unit, the dynamic obstacle detection module can detect mobile obstacles on the delivery path in real time and accurately, generate the location information and motion trajectory data of the obstacles, provide an important basis for UAV path planning, and ensure the safety and efficiency of the delivery mission.

[0093] The obstacle recognition unit specifically includes:

[0094] Image preprocessing: Grayscale the original image captured by the camera to reduce the color information of the image for subsequent edge extraction operations. The grayscale formula is: ,in, is the pixel value of the grayscale image, They are the pixel values ​​of the red, green and blue channels in the original image respectively;

[0095] Edge detection: Use the Canny edge detection algorithm to extract the edge of the preprocessed grayscale image to extract the outline of the obstacle. The gradient calculation formula of the Canny algorithm is: ,in, is the gradient value, and They are the pixel gradients of the image in the horizontal and vertical directions, respectively, indicating the intensity of edge changes in the image;

[0096] Shape feature extraction: Through the contour detection algorithm, the shape contour of the obstacle is extracted from the edge detection result, and its contour is represented as a series of point sets including ,The point set is used to form the two-dimensional boundary shape of the obstacle;

[0097] Position feature calculation: Based on the camera's intrinsic parameter matrix K and the extracted contour point set, the edge contour in the two-dimensional image is projected back to the three-dimensional space using inverse projection to calculate the position coordinates of the obstacle in the two-dimensional space. , the formula is:

[0098] ;

[0099] in, is the two-dimensional pixel coordinate in the image, is the coordinate of the obstacle in three-dimensional space, is the scale factor, is the camera internal parameter matrix;

[0100] Thermal signal verification: Combined with the thermal signal data collected by the infrared sensor, by setting the thermal signal threshold , to determine whether there is an obstacle; if the thermal signal value of a certain pixel satisfy: , it means there is an obstacle. By comparing the thermal signal and visual profile data, the existence and location of the obstacle can be further confirmed. By using computer vision algorithms combined with the thermal signal of the infrared sensor, the edge, shape and spatial position of the obstacle can be accurately extracted, and the existence of the obstacle can be verified by the thermal signal. This double verification method improves the accuracy of obstacle recognition and ensures that the drone can accurately identify and avoid obstacles in complex environments.

[0101] The hierarchical path planning module includes a data receiving unit, a height level planning unit, an obstacle avoidance unit, a no-fly zone processing unit and a path generation unit; wherein:

[0102] Data receiving unit: used to receive the building height, position, shape and distribution data from the 3D geographic information acquisition module, as well as the obstacle position information and motion trajectory data transmitted by the dynamic obstacle detection module. The 3D geographic information is used to build the basic spatial framework of the distribution area, and the obstacle information is used to assist the dynamic obstacle avoidance of the path;

[0103] Height level planning unit: Based on the three-dimensional geographic information provided by the data receiving unit, the delivery area is divided into multiple flight height levels. Each height level is divided according to the height distribution of buildings to ensure that the drone can flexibly switch between different height levels to avoid collision with buildings; specific flight height Determined by the following formula: ,in, is the flight altitude, is the height of the tallest building in the area, is the safe distance between the drone and the building, in meters;

[0104] Obstacle avoidance unit: Combines obstacle information, dynamically adjusts the flight path, and generates an adaptive obstacle avoidance path by analyzing the location information and movement trajectory of the obstacle. The calculation of the obstacle avoidance path is based on the shortest path algorithm to ensure that the drone can bypass obstacles. The path optimization formula is:

[0105] ,in, is the updated path;

[0106] No-fly zone processing unit: According to the marked no-fly zone data in the area, the no-fly zone is marked as a no-entry area, and the drone is prevented from entering the no-fly zone during path planning. The constraints of the no-fly zone are set as follows: ,in, is the current coordinate of the drone, The spatial coordinates of the no-fly zone to ensure that the flight path of the drone is away from the no-fly zone;

[0107] Path generation unit: It is used to combine the flight data generated by the altitude planning unit, obstacle avoidance unit and no-fly zone processing unit to generate a multi-level flight path. The generated path ensures that the UAV can avoid obstacles and no-fly zones in complex urban environments, and flexibly switch between multiple flight altitude levels to achieve the optimal flight route.

[0108] The path generation unit specifically includes:

[0109] Integrate flight data: receive the altitude data provided by the altitude planning unit, the obstacle avoidance path data provided by the obstacle avoidance unit, and the no-fly zone coordinate data provided by the no-fly zone processing unit. These data together constitute the flight constraints and reference path;

[0110] Path optimization: Based on the flight data provided by each unit, the path generation unit uses an optimization algorithm to generate the optimal path. By considering the flight altitude, obstacle location and no-fly zone boundaries, the shortest path algorithm is used to calculate the optimized path. , the optimization formula is:

[0111] ,in, is the optimal path, For each 3D coordinate point on the path, ensure that the path meets safety conditions in both height and horizontal directions. is the total number of points on the path;

[0112] Flight altitude adjustment: Based on the flight altitude provided by the altitude planning unit, the path generation unit dynamically adjusts the altitude level of the flight path according to the distribution and height of the buildings. The flight altitude calculation formula is: ,in, is the flight altitude, is the height of the tallest building in the covered area, is the safety distance in meters. This step ensures that the drone can switch freely between different levels to avoid collision;

[0113] Obstacle avoidance path generation: Combines the position and motion trajectory of the obstacle and generates a dynamic obstacle avoidance path by calculating the position change of the obstacle in the path , the obstacle avoidance path is calculated by dynamically adjusting the path points, and the formula is: ,in, is the adjusted obstacle avoidance path, For obstacles in time The spatial position of time, is the total number of points on the path;

[0114] No-fly zone processing: The path generation unit adjusts the flight path according to the coordinate data of the no-fly zone to ensure that the UAV flight path is away from the no-fly zone;

[0115] Final path generation: Integrate the adjusted obstacle avoidance path, flight altitude level, and path data after no-fly zone processing to generate the final multi-level flight path , the formula is: ,in, is the final generated path, is the adjusted obstacle avoidance path, is the flight altitude; by combining altitude level planning, obstacle avoidance and no-fly zone processing, a multi-level flight path suitable for complex urban environments can be generated. The final generated flight path can dynamically adjust the flight altitude, avoid obstacles and no-fly zones, and improve the safety and efficiency of delivery tasks.

[0116] The real-time environmental perception module includes a meteorological data acquisition unit, a data processing unit, and a risk assessment unit; wherein:

[0117] Meteorological data collection unit: A meteorological sensor installed on the drone that collects real-time data on meteorological conditions in the environment during flight, including wind speed (Unit: m / s), wind direction (unit: degrees) and air pressure (Unit: Pascal), weather condition data are continuously collected by meteorological sensors and transmitted to the data processing unit;

[0118] Data processing unit: receives real-time meteorological data from the meteorological data collection unit and compares it with the path point information in the multi-level flight path. Compare and analyze to determine whether meteorological conditions have an impact on each point in the path; the specific analysis formula is:

[0119] ,in, is the distance between the flight path point and the center of wind direction change, are the center coordinates of the area affected by wind speed and direction, Represents the drone in the path Points Axis coordinates, in meters, represent the horizontal position of the path point; Represents the drone in the path The Y-axis coordinate of each point, in meters, indicates the vertical position of the path point; Represents the drone in the path Points Axis coordinates, in meters, represent the position of the path point in the height direction (usually related to the ground height). Less than the preset safety distance , then the path point is considered to have potential risks;

[0120] Risk Assessment Unit: By comparing and analyzing the results, combined with the current wind speed and wind direction data, the potential risks on the flight path are assessed; if it is detected that any point on the path is less than the safety threshold from the center of the wind speed change area , then mark the point as a risk point; the specific judgment formula is:

[0121] , among which the risk points When , it means that there is a potential risk at this path point and the path needs to be adjusted; through the collaborative work of the meteorological data acquisition unit and the data processing unit, the real-time environment perception module can accurately collect meteorological data during the flight, and dynamically compare and analyze it with the flight path, so as to timely identify potential risks that affect flight safety and ensure that the UAV can make flight path adjustments according to current meteorological conditions, thereby improving the safety and reliability of delivery tasks.

[0122] The path adjustment module includes a risk input unit, a historical data analysis unit, an autonomous learning unit and a path output unit; wherein:

[0123] Risk input unit: receives risk data from the real-time environment perception module, including risk markers on the path points and the distance to the center of the wind speed variation area , and pass the path points marked as risks to the autonomous learning unit as trigger signals for path adjustment;

[0124] Historical data analysis unit: collects historical flight data, including weather conditions and path adjustment records encountered in previous flights, and predicts the changing pattern of the optimal flight path of the drone under similar weather conditions based on historical data; the path adjustment information in the historical data is specifically matched through the following formula:

[0125] ,

[0126] in, Indicates the adjusted difference between the current path and the historical path. are the coordinates of the waypoints in the historical flight, is the number of path points;

[0127] Autonomous learning unit: uses autonomous learning algorithms to integrate the current environmental data provided by the real-time environmental perception module and the historical flight data provided by the historical data analysis unit, and adjusts the flight path through multiple iterations;

[0128] Path output unit: outputs the adjusted new path to the UAV's flight control system to update flight instructions, ensuring that the UAV immediately optimizes the path after identifying risks, avoids dangerous areas and maintains flight safety and efficiency; by combining the risk identification and historical data analysis of the real-time environmental perception module, the path adjustment module can generate an optimized flight path in real time, ensuring that the UAV can flexibly respond to various meteorological changes in complex environments. The introduction of autonomous learning algorithms has greatly improved the accuracy and efficiency of path adjustment, enabling UAVs to complete delivery tasks more intelligently in dynamic environments.

[0129] The self-study units include:

[0130] Data input: Receive current environmental data from the real-time environmental perception module, including wind speed ,wind direction , air pressure and historical flight path adjustment data provided by the historical data analysis unit ;

[0131] Learning algorithm processing: Based on the received current environmental data and historical data, the reinforcement learning algorithm is used to optimize the flight path of the drone. The reinforcement learning algorithm uses the safety and path optimization of the drone as the reward function, and continuously updates the optimal path strategy through data feedback from multiple flight missions. The optimization objective function of reinforcement learning is: ,in, is the cumulative reward value of the future path, is the immediate reward for each time step, is the discount factor, is the number of future path points. The reinforcement learning algorithm maximizes the cumulative reward value To achieve optimal adjustment of the path;

[0132] Strategy update: During the learning process, the path adjustment strategy is continuously updated. Based on the feedback from reinforcement learning, the strategy update unit generates a new path adjustment direction. , the path is based on the current flight environment data and historical path adjustment experience to ensure that the drone can avoid risks and maintain flight efficiency. The formula for path adjustment is: ,in, For the new path, is the current path, is the risk assessment function, and is the path adjustment coefficient; risk assessment function The expression is: ,in, The risk factor is the output of the risk assessment function, ranging from 0 to 1, and is used to measure the risk of the drone's flight path; The distance between the current path point and the center of the wind speed change area, in meters , represents the distance from the flight path point to the center of the risk area, is the preset safety distance threshold. , it means the waypoint is in the danger zone; is a constant that controls the steepness of the curve and is used to adjust the sensitivity of the risk factor. Close to or less than , that is, when the path point is very close to the risk area, Close to 1, it indicates a higher risk and requires a larger adjustment of the path; keep away hour, Close to 0, it means the risk is small and there is no need to significantly adjust the path. The above-mentioned autonomous learning unit can optimize the flight path of the UAV by combining real-time environmental perception data and historical path adjustment data, and use reinforcement learning algorithms to continuously improve the UAV's ability to cope with complex environments. The unit can intelligently adjust the path according to the current environment and historical experience, so that the UAV can efficiently complete delivery tasks under different environmental conditions.

[0133] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0134] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. The intelligent logistics distribution path planning system of drones is characterized by: It includes a three-dimensional geographic information acquisition module, a dynamic obstacle detection module, a hierarchical path planning module, a real-time environment perception module and a path adjustment module; among which: 3D geographic information collection module: Deploy lidar and cameras on drones to collect 3D geographic information on building height, location, shape and distribution within the delivery area; Dynamic obstacle detection module: Combines the 3D geographic information provided by the 3D geographic information acquisition module and the dynamic information collected by the drone sensor to detect moving obstacles on the delivery path in real time, and generates the location information and motion trajectory data of the obstacles, which are then transmitted to the hierarchical path planning module; Hierarchical path planning module: Based on the 3D geographic information provided by the 3D geographic information acquisition module and the obstacle information transmitted by the dynamic obstacle detection module, combined with factors such as flight altitude, building distribution and no-fly zones, a multi-level flight path is generated; Real-time environmental perception module: Through the meteorological sensors deployed on the drone, real-time meteorological data, including wind speed, wind direction and air pressure, are collected, and the real-time meteorological condition data is compared and analyzed with the generated multi-level flight path information to identify potential risks that affect the flight path in the current environment; Path adjustment module: Based on the potential risks identified by the real-time environmental perception module, and combined with historical flight data and current environmental changes, the autonomous learning algorithm is used to adjust the flight path of the drone in real time to optimize the flight path.

2. The intelligent logistics distribution path planning system for drones according to claim 1 is characterized in that: The three-dimensional geographic information acquisition module includes a laser radar unit, a camera unit and a data processing unit; wherein: LiDAR unit: installed at the bottom of the drone, used to emit laser beams and receive signals reflected from the surface of buildings, and calculate the height and spatial position of buildings by measuring the signal return time; Camera unit: installed on the front of the drone, used to capture the external contour and shape information of the building and generate a two-dimensional image of the building; Data processing unit: electrically connected to the lidar unit and the camera unit, used to integrate the height and spatial position information generated by the lidar with the image data collected by the camera unit, and generate a three-dimensional geographic model of the delivery area.

3. The intelligent logistics distribution path planning system for drones according to claim 2 is characterized in that: The data processing unit specifically includes: Integrate height and spatial position information with 2D image data: Combine the building height and spatial coordinates generated by the lidar with the 2D image coordinates generated by the camera unit Match the pixels in each two-dimensional image Corresponding to a point in three-dimensional space , giving each pixel its real three-dimensional coordinates; Generate 3D point cloud data: Generate the 3D coordinate point corresponding to each pixel through the matched height and 2D image data , and the three-dimensional coordinate points form a three-dimensional point cloud data set of the building, each point represents a position on the surface of the building; Point cloud data optimization: based on 2D images taken from multiple angles, point cloud data can be stitched and redundant points can be eliminated; Constructing a 3D model of a building: Generate the 3D geometric shape of the building by converting the optimized point cloud data into a triangular mesh model. Specifically, the volume and appearance of the building are represented by connecting adjacent point cloud data points into a mesh. Generate 3D geographic model: Spatially fuse the 3D models of each building to generate a complete 3D geographic model of the distribution area.

4. The intelligent logistics distribution path planning system for drones according to claim 1 is characterized in that: The dynamic obstacle detection module includes an environment perception unit, an obstacle recognition unit and a motion trajectory prediction unit; wherein: Environmental perception unit: used to receive the three-dimensional geographic information of the delivery area from the three-dimensional geographic information acquisition module, and combine with the camera and infrared sensor deployed on the drone to capture the real-time image and thermal signal of dynamic objects on the delivery path; Obstacle recognition unit: It identifies moving obstacles in the delivery path by analyzing the images and infrared data collected by the environmental perception unit. It uses computer vision algorithms to process the images collected by the camera, extracts the edge, shape and position features of the obstacle, and verifies whether the obstacle exists by combining the thermal signal provided by the infrared sensor. Motion trajectory prediction unit: Based on the obstacle position information provided by the obstacle recognition unit and combined with continuous image frames, the motion trajectory of the obstacle is calculated. The formula is: ； in, is the speed of the obstacle; Indicates the obstacle at time The three-dimensional space coordinates of the time, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; Indicates the obstacle at time The three-dimensional space coordinates of the time, where For time The X-axis coordinate of the obstacle at the moment; For time The Y-axis coordinate of the obstacle at the moment; For time Time obstacles Axis coordinates; They are all time points, used to describe the location information of obstacles at different times. .

5. The intelligent logistics distribution path planning system for drones according to claim 4 is characterized in that: The obstacle recognition unit specifically includes: Image preprocessing: grayscale the original image captured by the camera to reduce the color information of the image; Edge detection: Use the Canny edge detection algorithm to extract the edges of the preprocessed grayscale image to extract the outline of the obstacle; Shape feature extraction: Through the contour detection algorithm, the shape contour of the obstacle is extracted from the edge detection result, and its contour is represented as a series of point sets including ,The point set is used to form the two-dimensional boundary shape of the obstacle; Position feature calculation: Based on the camera's intrinsic parameter matrix K and the extracted contour point set, the edge contour in the two-dimensional image is projected back to the three-dimensional space using inverse projection to calculate the position coordinates of the obstacle in the two-dimensional space. ; Thermal signal verification: Combined with the thermal signal data collected by the infrared sensor, by setting the thermal signal threshold , to determine whether there is an obstacle; if the thermal signal value of a certain pixel satisfy: , it means there is an obstacle.

6. The intelligent logistics distribution path planning system for drones according to claim 1 is characterized in that: The hierarchical path planning module includes a data receiving unit, a height level planning unit, an obstacle avoidance unit, a no-fly zone processing unit and a path generation unit; wherein: Data receiving unit: used to receive the building height, position, shape and distribution data from the three-dimensional geographic information acquisition module, as well as the obstacle position information and motion trajectory data transmitted by the dynamic obstacle detection module. The three-dimensional geographic information is used to construct the basic spatial framework of the distribution area, and the obstacle information is used to assist the dynamic obstacle avoidance of the path; Altitude planning unit: Based on the three-dimensional geographic information provided by the data receiving unit, the delivery area is divided into multiple flight altitude levels. Each altitude level is divided according to the height distribution of buildings to ensure that the drone can flexibly switch between different altitude levels to avoid collisions with buildings. Obstacle avoidance unit: dynamically adjusts the flight path based on obstacle information, generates an adaptive obstacle avoidance path by analyzing the location information and movement trajectory of the obstacle; the calculation of the obstacle avoidance path is based on the shortest path algorithm to ensure that the drone can bypass the obstacle; No-fly zone processing unit: According to the marked no-fly zone data in the area, the no-fly zone is marked as a no-entry area, and the drone is prevented from entering the no-fly zone during path planning. The constraints of the no-fly zone are set as follows: ,in, is the current coordinate of the drone, is the spatial coordinate of the no-fly zone; Path generation unit: used to generate a multi-level flight path by combining the flight data generated by the altitude level planning unit, obstacle avoidance unit and no-fly zone processing unit.

7. The intelligent logistics distribution path planning system for unmanned aerial vehicles according to claim 6 is characterized in that: The path generation unit specifically includes: Integrate flight data: receive altitude data provided by the altitude planning unit, obstacle avoidance path data provided by the obstacle avoidance unit, and no-fly zone coordinate data provided by the no-fly zone processing unit; Path optimization: Based on the flight data provided by each unit, the path generation unit uses an optimization algorithm to generate the optimal path. By considering the flight altitude, obstacle location and no-fly zone boundaries, the shortest path algorithm is used to calculate the optimized path. ; Flight altitude adjustment: Based on the flight altitude provided by the altitude planning unit, the path generation unit dynamically adjusts the altitude level of the flight path according to the distribution and height of the buildings; Obstacle avoidance path generation: Combines the position and motion trajectory of the obstacle and generates a dynamic obstacle avoidance path by calculating the position change of the obstacle in the path ; No-fly zone processing: The path generation unit adjusts the flight path according to the coordinate data of the no-fly zone to ensure that the UAV flight path is away from the no-fly zone; Final path generation: Integrate the adjusted obstacle avoidance path, flight altitude level, and path data after no-fly zone processing to generate the final multi-level flight path , the formula is: = ,in, is the final generated path, is the adjusted obstacle avoidance path, For the flight altitude.

8. The intelligent logistics distribution path planning system for drones according to claim 1 is characterized in that: The real-time environment perception module includes a meteorological data acquisition unit, a data processing unit and a risk assessment unit; wherein: Meteorological data collection unit: A meteorological sensor installed on the drone that collects real-time data on meteorological conditions in the environment during flight, including wind speed ,wind direction and air pressure , meteorological condition data are continuously collected by meteorological sensors and transmitted to the data processing unit; Data processing unit: Receives real-time meteorological data provided by the meteorological data acquisition unit, and compares and analyzes it with the path point information in the multi-level flight path to determine whether the meteorological conditions have an impact on each point in the path; the specific analysis formula is: ,in, is the distance between the flight path point and the center of wind direction change, are the center coordinates of the area affected by wind speed and direction, Represents the drone in the path Points Axis coordinates; Represents the drone in the path The Y-axis coordinate of a point; Represents the drone in the path Points Axis coordinates, in meters; Risk Assessment Unit: By comparing and analyzing the results, combined with the current wind speed and wind direction data, the potential risks on the flight path are assessed; if it is detected that any point on the path is less than the safety threshold from the center of the wind speed change area , then mark the point as a risk point.

9. The intelligent logistics distribution path planning system for drones according to claim 8, characterized in that: The path adjustment module includes a risk input unit, a historical data analysis unit, an autonomous learning unit and a path output unit; wherein: Risk input unit: receives risk data from the real-time environment perception module, including risk markers on the path points and the distance to the center of the wind speed variation area , and pass the path points marked as risks to the autonomous learning unit as trigger signals for path adjustment; Historical data analysis unit: collects historical flight data, including weather conditions and path adjustment records encountered in previous flights, and predicts the changing pattern of the optimal flight path of the drone under similar weather conditions based on historical data; Autonomous learning unit: uses autonomous learning algorithms to integrate the current environmental data provided by the real-time environmental perception module and the historical flight data provided by the historical data analysis unit, and adjusts the flight path through multiple iterations; Path output unit: outputs the adjusted new path to the UAV's flight control system for updating flight instructions.

10. The intelligent logistics distribution path planning system for drones according to claim 9, characterized in that: The autonomous learning unit specifically includes: Data input: Receive current environmental data from the real-time environmental perception module, including wind speed ,wind direction , air pressure and historical flight path adjustment data provided by the historical data analysis unit ; Learning algorithm processing: Based on the received current environment data and historical data, the reinforcement learning algorithm is used to optimize the flight path of the drone. The reinforcement learning algorithm uses the safety and path optimization of the drone as the reward function, and continuously updates the optimal path strategy through data feedback from multiple flight missions. The optimization objective function of reinforcement learning is: ,in, is the cumulative reward value of the future path, is the immediate reward for each time step, is the discount factor, is the number of future path points; Strategy update: During the learning process, the path adjustment strategy is continuously updated. Based on the feedback from reinforcement learning, the strategy update unit generates a new path adjustment direction. , the formula for path adjustment is: ,in, For the new path, is the current path, is the risk assessment function, and is the path adjustment factor.

Images