Indoor unmanned aerial vehicle flight path planning processing method and system

Abstract

The invention discloses an indoor unmanned aerial vehicle flight path planning processing method and system, and the method comprises the steps: scanning an indoor environment through a laser radar, and obtaining real-time obstacle information; the indoor environment is scanned through laser radar equipment, so that point cloud information and obstacle position information of a plurality of obstacles in the environment can be obtained; dividing the grid map into a plurality of levels of subareas according to the density of the obstacles in the grids and the probability value of the occupied area; generating an initial flight path by adopting an improved fast expansion random tree algorithm based on the partition regions of the multiple levels; smoothing the initial flight path to obtain an optimized flight path; an improved fast expansion random tree algorithm is adopted to generate an initial path, and compared with a traditional algorithm, the algorithm has higher calculation speed and better path quality.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle flight control, and in particular to a method and system for indoor unmanned aerial vehicle flight path planning. Background Art

[0002] The development of indoor drone obstacle avoidance technology has experienced a transformation from simple mechanical obstacle avoidance to complex intelligent obstacle avoidance. Early obstacle avoidance technologies mainly relied on mechanical sensors, such as infrared sensors and ultrasonic sensors, which can provide limited obstacle avoidance information, but have limited detection range and accuracy. With the development of computer vision technology, image recognition-based obstacle avoidance technology has begun to emerge, which can provide richer environmental information, but has large computational complexity and high hardware requirements.

[0003] Specifically, indoor drone flight control and obstacle avoidance technology is also evolving, driving the continuous innovation and progress of indoor drone applications. In practical applications, it is often necessary to use a variety of technologies in combination according to specific scenarios and needs to achieve safe and efficient indoor drone flight control and obstacle avoidance functions.

[0004] The obstacle avoidance method in the prior art is flight obstacle detection and obstacle avoidance control based on image recognition. This control method has too much computational complexity, resulting in a long hovering time. The computing power greatly affects the efficiency of indoor drone transportation or indoor drone safety operations. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for indoor UAV flight path planning and processing, which solves the above-mentioned technical problems pointed out in the prior art.

[0006] The present invention proposes a method for indoor UAV flight path planning and processing, comprising the following steps:

[0007] Acquire three-dimensional map data of the indoor environment; the three-dimensional map data is three-dimensional map data pre-constructed in a database;

[0008] Use laser radar to scan the indoor environment and obtain real-time obstacle information; that is, scan the indoor environment through laser radar equipment, so as to obtain multiple obstacle point cloud information and obstacle location information in the environment;

[0009] A grid map is constructed based on three-dimensional map data and real-time obstacle information to obtain the density and occupancy probability value of obstacles in each grid; the grid map is divided into multiple levels of partition areas according to the density and occupancy probability value of obstacles in the grid;

[0010] Based on multiple levels of partitioned areas, an improved fast-expanding random tree algorithm is used to generate the initial flight path;

[0011] Smoothing the initial flight path to obtain an optimized flight path;

[0012] Control the drone to fly according to the optimized flight path.

[0013] Preferably, after controlling the drone to fly according to the optimized flight path, the method further includes:

[0014] Obstacle images on the flight path are collected, image recognition is performed based on the obstacle images to determine the current obstacle type, dangerous obstacles are screened according to the identified current obstacle type, dangerous obstacle location nodes are marked, and the flight path is readjusted.

[0015] Preferably, after controlling the drone to fly according to the optimized flight path, the method further includes:

[0016] Obstacle images on the flight path are collected, image recognition is performed based on the obstacle images to determine the current obstacle type, dangerous obstacles are screened according to the identified current obstacle type, and dangerous obstacle location nodes are marked.

[0017] Preferably, collecting obstacle images on the flight path, performing image recognition based on the obstacle images to determine the current obstacle type, screening dangerous obstacles according to the identified current obstacle type, and marking dangerous obstacle location nodes, specifically including:

[0018] During the flight, set the acquisition frequency and use the camera on the drone to regularly capture images of obstacles on the path;

[0019] The collected images are preprocessed and analyzed using a deep learning visual algorithm to identify and determine whether the current obstacle type is a dangerous obstacle. If so, the identified dangerous obstacle is marked and its location information is recorded.

[0020] Preferably, the grid map is constructed based on the three-dimensional map data and the real-time obstacle information, including:

[0021] Discretize the three-dimensional map data into three-dimensional grids;

[0022] Mapping real-time obstacle information to the corresponding three-dimensional grid;

[0023] Determine the density of obstacles and the occupancy probability value of the grid for each grid.

[0024] Preferably, obtaining the density and occupancy probability value of obstacles in each grid includes:

[0025] Obtain the number of obstacles in each grid in the three-dimensional grid and calculate the density of obstacles in the grid; the density of obstacles is the number of independent obstacles in a single grid;

[0026] Determine the occupancy status of the grid based on the obstacle density;

[0027] At the same time, the occupied surface probability value of the obstacle in the current grid is calculated, and the occupied surface probability value of the obstacle is the ratio of the occupied obstacle area in the grid to the current grid area.

[0028] Preferably, the grid map is divided into multiple levels of partition areas according to the density of obstacles in the grid and the probability value of occupied surface, specifically including:

[0029] For each grid, a score is calculated based on the obstacle density and the occupancy probability value;

[0030] Score = w1*obstacle density+w2*occupancy probability value; where w1 and w2 are the preset first weight and second weight;

[0031] Define multiple thresholds A1 and A2. The area with a score less than or equal to A1 is a low-risk area, the area with a score greater than or equal to A2 is a high-risk area, and the area with a score between A1 and A2 is a medium-risk area.

[0032] Traverse the entire current grid map, calculate the score for each grid in the grid map, and divide the risk level according to the threshold;

[0033] Adjacent grids of the same risk level are clustered to form continuous risk areas.

[0034] Preferably, based on multiple levels of partitioned areas, an improved fast-expanding random tree algorithm is used to generate the initial flight path, specifically including:

[0035] Perform node sampling on the current multiple levels of partitioned areas. When sampling, assign different sampling frequencies to areas of different risk levels to reduce the sampling probability of high-risk areas; the sampling frequency gradually decreases from low-risk areas, medium-risk areas to high-risk areas;

[0036] Set the starting point and the end point, run the improved RRT algorithm, and generate the initial flight path; when executing, the node at the starting point is used as the initial candidate node, and the distance between the current candidate node and multiple existing sampled tree nodes is calculated within the preset distance range;

[0037] Within the preset distance range, select n tree nodes that are closer to the current candidate node as parent nodes, and use the local planner to connect the parent nodes and candidate nodes;

[0038] Check whether the connection path intersects with obstacles;

[0039] If they do not intersect, the candidate node is added to the tree;

[0040] Repeat the above steps until a feasible path is found or the maximum number of iterations is reached and the generation of the initial flight path is completed.

[0041] Preferably, after the generation of the initial flight path is completed, the method further includes:

[0042] The number of current initial flight paths is judged. If there are multiple current initial flight paths generated by the improved rapid expansion random tree algorithm, the number of areas of each risk level passed by the multiple current initial flight paths is calculated, and the initial flight path with the largest number of areas passing through the low-risk area is selected as the final target flight path.

[0043] Preferably, the process of smoothing the initial flight path includes:

[0044] Use cubic spline interpolation to interpolate path points;

[0045] Then the interpolated path points are subjected to a local optimization algorithm to adjust the path curvature.

[0046] Preferably, the performing of a local optimization algorithm to adjust the path curvature specifically includes:

[0047] Calculate the curvature of each point on the path;

[0048] For points where the curvature exceeds the threshold, the following optimization formula is applied:

[0049] P'i=Pi+λ*SPi-1+Pi+1-2Pi+μ*SPi+2+Pi-2-2Pi);

[0050] Among them, P'i is the optimized point coordinate, Pi is the current point coordinate, Pi-1, Pi+1, Pi-2 and Pi+2 are the adjacent point coordinates, λ and μ are weight coefficients;

[0051] The above steps are iterated until the curvature meets the requirements or the maximum number of iterations is reached.

[0052] The present invention provides an indoor UAV flight path planning and processing system, comprising:

[0053] A three-dimensional map data acquisition module is used to acquire three-dimensional map data of an indoor environment; the three-dimensional map data is three-dimensional map data pre-constructed in a database;

[0054] The laser radar scanning module is used to use the laser radar to scan the indoor environment and obtain real-time obstacle information; that is, the indoor environment is scanned by the laser radar device, so as to obtain multiple obstacle point cloud information and obstacle position information in the environment;

[0055] A grid map construction module is used to construct a grid map based on three-dimensional map data and real-time obstacle information, obtain the density of obstacles and the probability value of occupied surfaces in each grid; divide the grid map into multiple levels of partition areas according to the density of obstacles and the probability value of occupied surfaces in the grid;

[0056] A path generation module is used to generate an initial flight path based on multiple levels of partitioned areas using an improved rapidly expanding random tree algorithm;

[0057] The path optimization module smoothes the initial flight path to obtain the optimized flight path;

[0058] The flight control module is used to control the drone to fly according to the optimized flight path.

[0059] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0060] By analyzing the above-mentioned indoor UAV flight path planning processing method and system provided by the present invention, it can be known that in specific applications, the three-dimensional map data of the indoor environment is obtained; the three-dimensional map data is pre-constructed three-dimensional map data in the database; by combining the pre-constructed three-dimensional map data (S1) and the obstacle information (S2) scanned in real time by the laser radar, the method can obtain more accurate and timely environmental information. This helps to generate a more accurate flight path and can adapt to the dynamic changes of the environment.

[0061] The indoor environment is scanned by laser radar to obtain real-time obstacle information; that is, the indoor environment is scanned by laser radar equipment, so that multiple obstacle point cloud information and obstacle location information in the environment can be obtained; a grid map is constructed based on three-dimensional map data and real-time obstacle information to obtain the density and occupancy probability value of obstacles in each grid; the grid map is divided into multiple levels of partitioned areas according to the density and occupancy probability value of obstacles in the grid; based on the multiple levels of partitioned areas, an improved fast expansion random tree algorithm is used to generate an initial flight path;

[0062] The grid map and obstacle density / occupancy probability method can be used to describe the obstacle distribution in the environment in more detail. The study found that this hierarchical area division helps drones better identify and avoid areas with different risk levels. Based on this technology, an improved rapidly expanding random tree algorithm is used to generate the initial path, which has faster calculation speed and better path quality than the traditional algorithm. The initial flight path is smoothed to obtain the optimized flight path; the drone is controlled to fly according to the optimized flight path.

[0063] The above-mentioned indoor UAV flight path planning and processing method and system provided by the present invention are specially designed for indoor environments, taking into account the characteristics of indoor environments (such as narrow spaces, multiple obstacles, etc.), and are therefore particularly suitable for application in complex indoor environments, while also embodying the system's high autonomy and intelligence. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of the main flow of an indoor UAV flight path planning and processing method according to the first embodiment of the present invention;

[0065] Figure 2 A schematic diagram of a specific implementation process of a method for indoor UAV flight path planning and processing according to embodiment 1 of the present invention;

[0066] Figure 3 This is another specific implementation process diagram of an indoor UAV flight path planning and processing method according to the first embodiment of the present invention;

[0067] Figure 4 This is a schematic diagram of another specific implementation process of an indoor UAV flight path planning and processing method according to the first embodiment of the present invention;

[0068] Figure 5 This is a schematic diagram of the overall architecture of an indoor UAV flight path planning and processing system according to Embodiment 2 of the present invention.

[0069] Reference numerals: three-dimensional map data acquisition module 10 , laser radar scanning module 20 , grid map construction module 30 , path generation module 40 , path optimization module 50 , flight control module 60 . DETAILED DESCRIPTION

[0070] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0072] Embodiment 1

[0073] like Figure 1 As shown, the first embodiment of the present invention provides a method for indoor UAV flight path planning and processing, including the following operation steps:

[0074] S1. Acquire three-dimensional map data of an indoor environment; the three-dimensional map data is three-dimensional map data pre-constructed in a database;

[0075] S2. Use laser radar to scan the indoor environment and obtain real-time obstacle information; that is, scan the indoor environment through the laser radar device, so as to obtain multiple obstacle point cloud information and obstacle position information in the environment;

[0076] S3, constructing a grid map based on the three-dimensional map data and real-time obstacle information, obtaining the density of obstacles and the occupancy probability value in each grid; dividing the grid map into multiple levels of partition areas according to the density of obstacles and the occupancy probability value in the grid;

[0077] S4, based on multiple levels of partitioned areas, an improved fast-expanding random tree algorithm is used to generate the initial flight path;

[0078] S5, smoothing the initial flight path to obtain an optimized flight path;

[0079] S6. Control the drone to fly according to the optimized flight path.

[0080] By analyzing the above-mentioned indoor UAV flight path planning processing method and system provided by the present invention, it can be known that in specific applications, the three-dimensional map data of the indoor environment is obtained; the three-dimensional map data is pre-constructed three-dimensional map data in the database; by combining the pre-constructed three-dimensional map data (S1) and the obstacle information (S2) scanned in real time by the laser radar, the method can obtain more accurate and timely environmental information. This helps to generate a more accurate flight path and can adapt to the dynamic changes of the environment.

[0081] The indoor environment is scanned by laser radar to obtain real-time obstacle information; that is, the indoor environment is scanned by laser radar equipment, so that multiple obstacle point cloud information and obstacle location information in the environment can be obtained; a grid map is constructed based on three-dimensional map data and real-time obstacle information to obtain the density and occupancy probability value of obstacles in each grid; the grid map is divided into multiple levels of partitioned areas according to the density and occupancy probability value of obstacles in the grid; based on the multiple levels of partitioned areas, an improved fast expansion random tree algorithm is used to generate an initial flight path;

[0082] The grid map and obstacle density / occupancy probability method can be used to describe the obstacle distribution in the environment in more detail. The study found that this hierarchical area division helps drones better identify and avoid areas with different risk levels. Based on this technology, an improved rapidly expanding random tree algorithm is used to generate the initial path, which has faster calculation speed and better path quality than the traditional algorithm. The initial flight path is smoothed to obtain the optimized flight path; the drone is controlled to fly according to the optimized flight path.

[0083] The above-mentioned indoor UAV flight path planning and processing method and system provided by the present invention are specially designed for indoor environments, taking into account the characteristics of indoor environments (such as narrow spaces, multiple obstacles, etc.), and are therefore particularly suitable for application in complex indoor environments, while also embodying the system's high autonomy and intelligence.

[0084] Preferably, after controlling the drone to fly according to the optimized flight path, the method further includes:

[0085] Obstacle images on the flight path are collected, image recognition is performed based on the obstacle images to determine the current obstacle type, dangerous obstacles are screened according to the identified current obstacle type, and dangerous obstacle location nodes are marked.

[0086] Preferably, collecting obstacle images on the flight path, performing image recognition based on the obstacle images to determine the current obstacle type, screening dangerous obstacles according to the identified current obstacle type, and marking dangerous obstacle location nodes, specifically including:

[0087] During the flight, set the acquisition frequency and use the camera on the drone to regularly capture images of obstacles on the path;

[0088] Obtain the preset standard shape of the dangerous obstacle, pre-process the collected image, use the deep learning vision algorithm to analyze and extract the pre-processed image to obtain the outer contour shape of the currently collected image, compare it with the standard shape of the dangerous obstacle to determine their similarity (if the calculated similarity is greater than the standard threshold, it is considered a successful match), identify and determine whether the current obstacle type is a dangerous obstacle, if so, mark the identified dangerous obstacle, and record its location information (such as coordinates, relative height, etc.).

[0089] See also Figure 2 The step of constructing a grid map based on the three-dimensional map data and the real-time obstacle information includes:

[0090] S31, discretizing the three-dimensional map data into three-dimensional grids;

[0091] S32, mapping the real-time obstacle information into a corresponding three-dimensional grid;

[0092] S33. Determine the density of obstacles and the occupancy probability value of the grid for each grid.

[0093] See also Figure 3 , the obtaining of the density and occupancy probability value of obstacles in each grid includes:

[0094] S331, obtaining the number of obstacles in each grid in the three-dimensional grid, and calculating the density of obstacles in the grid; the density of obstacles is the number of independent obstacles in a single grid;

[0095] S332, determining the occupancy state of the grid according to the obstacle density;

[0096] S333. Simultaneously calculate an occupied surface probability value of an obstacle in the current grid, where the occupied surface probability value of the obstacle is a ratio of an occupied obstacle area in the grid to a current grid area.

[0097] The definitions of occupancy state and occupancy probability values ​​are as follows: Occupancy State: This is a discrete concept, usually divided into three states: occupied, idle, and unknown. Occupancy: Indicates that there are obstacles in the grid. Idle: Indicates that there are no obstacles in the grid. Unknown: Indicates that it is impossible to determine whether the grid is occupied.

[0098] Occupancy probability value: This is a continuous concept, which indicates the probability of the ratio of the occupied obstacle area in the grid to the current grid area. It is a value between 0 and 1. The occupancy state is a discrete classification result. The occupancy probability value is a continuous numerical estimate. The grid map is divided into multiple levels of areas according to the obstacle density and the occupancy probability value, such as high-risk area, medium-risk area and low-risk area.

[0099] Preferably, the density of obstacles in the grid and the occupancy probability value divide the grid map into multiple levels of partition areas, specifically including:

[0100] For each grid, a score is calculated based on the obstacle density and the occupancy probability value;

[0101] Score = w1*obstacle density+w2*occupancy probability value; where w1 and w2 are the preset first weight and second weight;

[0102] Define multiple thresholds A1 and A2. The area with a score less than or equal to A1 is a low-risk area, the area with a score greater than or equal to A2 is a high-risk area, and the area with a score between A1 and A2 is a medium-risk area.

[0103] Traverse the entire current grid map, calculate the score for each grid in the grid map, and divide the risk level according to the threshold;

[0104] Adjacent grids of the same risk level are clustered to form continuous risk areas.

[0105] See also Figure 4 , based on multiple levels of partitioned areas, an improved rapidly expanding random tree algorithm is used to generate the initial flight path:

[0106] S41, performing node sampling on the current multiple levels of partitioned areas. During sampling, different sampling frequencies are assigned to areas of different risk levels to reduce the sampling probability of high-risk areas; wherein the sampling frequency gradually decreases from low-risk areas, medium-risk areas to high-risk areas;

[0107] S42, setting the starting point and the end point, running the improved RRT algorithm, and generating an initial flight path; when specifically executed, starting from the node at the starting point as the initial candidate node, calculating the distance between the current candidate node and the multiple tree nodes currently sampled within a preset distance range;

[0108] S43, within a preset distance range, select n tree nodes that are closer to the current candidate node as parent nodes, and use a local planner to connect the parent nodes and the candidate nodes;

[0109] S44, checking whether the connection path intersects with an obstacle;

[0110] S45. If they do not intersect, then add the candidate node to the tree;

[0111] S46. Repeat steps S41-S45 until a feasible path is found or the maximum number of iterations is reached and the generation of the initial flight path is completed.

[0112] Specific implementation case: a 10x10 indoor environment grid map, each grid size is 0.5mx 0.5m. Scoring function: Score = 0.6*obstacle density + 0.4*occupancy probability value;

[0113] Set thresholds: A1 = 0.3 (the boundary between low risk and medium risk); A2 = 0.7 (the boundary between medium risk and high risk); traverse the grid and calculate the score:

[0114] Grid (2, 3): obstacle density = 0.2, occupied surface probability = 0.15; Score (2, 3) = 0.6*0.2+0.4*0.15 = 0.18; Risk level: low risk.

[0115] Grid (5, 6): obstacle density = 0.5, occupied surface probability = 0.6; Score (5, 6) = 0.6*0.5+0.4*0.6 = 0.54; Risk level: medium risk.

[0116] Grid (8, 7): obstacle density = 0.9, occupied surface probability = 0.8; Score (8, 7) = 0.6*0.9+0.4*0.8 = 0.86; Risk level: high risk.

[0117] Regional clustering: adjacent rasters of the same risk level are combined into regions.

[0118] Preferably, after completing the generation of the initial flight path, the number of current initial flight paths is also determined. If there are multiple current initial flight paths generated by the improved rapidly expanding random tree algorithm, the number of areas of each risk level passed by the multiple current initial flight paths is calculated, and the initial flight path with the largest number of low-risk areas is selected as the final target flight path.

[0119] It should be noted that, considering the risk level of the path, paths passing through low-risk areas are given priority.

[0120] By assigning different sampling frequencies to areas with different risk levels, the probability of sampling in high-risk areas is reduced. This strategy effectively guides the path to avoid high-risk areas, thereby greatly improving the safety of the generated path. At the same time, the technical solution adopts a risk level differentiated sampling strategy that not only improves safety, but also improves the efficiency of path planning. Because the algorithm tends to perform more sampling in low-risk areas, this usually finds a feasible path faster.

[0121] In the specific implementation process, multiple tree nodes with close distances are selected as parent nodes instead of only selecting the nearest node. This strategy can improve the diversity and flexibility of the path and help find a better path.

[0122] The above technical solution adopted in the embodiment of the present invention checks whether the connection path intersects with obstacles when adding new nodes (S44-S45), which ensures that the generated path is feasible and avoids the situation where the drone passes through obstacles. The algorithm continuously expands the tree structure through repeated iterations (S46). This method can achieve a good balance between global path exploration and local path optimization. Therefore, it is said that it balances global exploration and local optimization. By setting the maximum number of iterations (S46), the algorithm avoids the risk of falling into an infinite loop and improves the robustness and reliability of the system.

[0123] Preferably, the process of smoothing the initial flight path in step S5 includes:

[0124] S51, interpolating the path points using a cubic spline interpolation method;

[0125] S52, then the interpolated path points are subjected to a local optimization algorithm to adjust the path curvature.

[0126] Preferably, the process of adjusting the path curvature by using a local optimization algorithm in step S52 includes:

[0127] S521, calculating the curvature of each point on the path;

[0128] S522. For points whose curvature exceeds the threshold, apply the following optimization formula:

[0129] P'i=Pi+λ*SPi-1+Pi+1-2Pi+μ*SPi+2+Pi-2-2Pi);

[0130] Among them, P'i is the optimized point coordinate, Pi is the current point coordinate, Pi-1, Pi+1, Pi-2 and Pi+2 are the adjacent point coordinates, λ and μ are weight coefficients;

[0131] S523. Iterate steps S521 and S522 until the curvature meets the requirement or the maximum number of iterations is reached.

[0132] By analyzing the above technical solutions, it can be seen that the above S521-S523 technical solutions adopted in the embodiment of the present invention can generate a smooth curve between the original path points by using the cubic spline interpolation method. This method can significantly improve the continuity and smoothness of the path, making the flight trajectory of the drone smoother and more natural. A smooth flight path can reduce the acceleration changes of the drone during flight, thereby optimizing the flight dynamics performance. This can not only improve the stability of the flight, but also reduce energy consumption and extend the flight time.

[0133] The cubic spline interpolation method can generate more intermediate path points while maintaining the original key path points. This increases the accuracy of the path, allowing the drone to fly more accurately along the planned path.

[0134] Finally, it is necessary to improve the local characteristics of the path; that is, the path curvature is adjusted by a local optimization algorithm (S52), which can further improve the local characteristics of the path. This optimization can eliminate sharp turns or unreasonable curvature changes in the path, making the path more reasonable and executable.

[0135] Obviously, a smooth path with reasonable curvature can reduce the violent movement of the drone during flight, thereby reducing the risk of loss of control or collision and improving flight safety.

[0136] Embodiment 2

[0137] See also Figure 5 The present invention provides an indoor UAV flight path planning and processing system, including: a three-dimensional map data acquisition module 10, a laser radar scanning module 20, a grid map construction module 30, a path generation module 40, a path optimization module 50, and a flight control module 60;

[0138] The three-dimensional map data acquisition module 10 is used to acquire the three-dimensional map data of the indoor environment; the three-dimensional map data is the three-dimensional map data pre-constructed in the database;

[0139] The laser radar scanning module 20 is used to scan the indoor environment using the laser radar to obtain real-time obstacle information; that is, the indoor environment is scanned by the laser radar device, so that multiple obstacle point cloud information and obstacle position information in the environment can be obtained;

[0140] The grid map construction module 30 is used to construct a grid map based on the three-dimensional map data and the real-time obstacle information, obtain the density of obstacles and the occupancy probability value in each grid; divide the grid map into multiple levels of partition areas according to the density of obstacles and the occupancy probability value in the grid;

[0141] A path generation module 40, for generating an initial flight path based on multiple levels of partitioned areas using an improved fast expanding random tree algorithm;

[0142] A path optimization module 50 performs smoothing on the initial flight path to obtain an optimized flight path;

[0143] The flight control module 60 is used to control the UAV to fly according to the optimized flight path.

[0144] In summary, the above-mentioned indoor UAV flight path planning and processing method and system provided by the present invention are specially designed for indoor environments, taking into account the characteristics of indoor environments (such as narrow spaces, multiple obstacles, etc.), and are therefore particularly suitable for application in complex indoor environments, while also reflecting the system's high autonomy and intelligence.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace part or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for indoor UAV flight path planning and processing, characterized in that: The steps are as follows: Acquire three-dimensional map data of the indoor environment; the three-dimensional map data is three-dimensional map data pre-constructed in a database; Use LiDAR to scan the indoor environment and obtain real-time obstacle information; That is, by scanning the indoor environment through the laser radar equipment, the point cloud information and obstacle location information of multiple obstacles in the environment can be obtained; A grid map is constructed based on three-dimensional map data and real-time obstacle information to obtain the density and occupancy probability value of obstacles in each grid; the grid map is divided into multiple levels of partition areas according to the density and occupancy probability value of obstacles in the grid; Based on multiple levels of partitioned areas, an improved fast-expanding random tree algorithm is used to generate the initial flight path; Smoothing the initial flight path to obtain an optimized flight path; Control the drone to fly according to the optimized flight path.

2. The indoor UAV flight path planning and processing method according to claim 1 is characterized in that: After controlling the drone to fly according to the optimized flight path, it also includes: Obstacle images on the flight path are collected, image recognition is performed based on the obstacle images to determine the current obstacle type, dangerous obstacles are screened according to the identified current obstacle type, and dangerous obstacle location nodes are marked.

3. The indoor UAV flight path planning and processing method according to claim 2 is characterized in that: Collecting obstacle images on the flight path, performing image recognition based on the obstacle images to determine the current obstacle type, screening dangerous obstacles according to the identified current obstacle type, and marking dangerous obstacle location nodes, specifically including: During the flight, set the acquisition frequency and use the camera on the drone to regularly capture images of obstacles on the path; The collected images are preprocessed and analyzed using a deep learning visual algorithm to identify and determine whether the current obstacle type is a dangerous obstacle. If so, the identified dangerous obstacle is marked and its location information is recorded.

4. The indoor UAV flight path planning and processing method according to claim 1 is characterized in that: The grid map is constructed based on the three-dimensional map data and the real-time obstacle information, including: Discretize the three-dimensional map data into three-dimensional grids; Mapping real-time obstacle information to the corresponding three-dimensional grid; Determine the density of obstacles and the occupancy probability value of the grid for each grid.

5. The indoor UAV flight path planning and processing method according to claim 4 is characterized in that: Obtaining the density and occupancy probability value of obstacles in each grid includes: Obtain the number of obstacles in each grid in the three-dimensional grid and calculate the density of obstacles in the grid; the density of obstacles is the number of independent obstacles in a single grid; Determine the occupancy status of the grid based on the obstacle density; At the same time, the occupied surface probability value of the obstacle in the current grid is calculated, and the occupied surface probability value of the obstacle is the ratio of the occupied obstacle area in the grid to the current grid area.

6. The indoor UAV flight path planning and processing method according to claim 5 is characterized in that: The grid map is divided into multiple levels of partition areas according to the density of obstacles and the probability value of occupied surfaces within the grid, including: For each grid, a score is calculated based on the obstacle density and the occupancy probability value; Score = w1*obstacle density+w2*occupancy probability value; where w1 and w2 are the preset first weight and second weight; Define multiple thresholds A1 and A2. The area with a score less than or equal to A1 is a low-risk area, the area with a score greater than or equal to A2 is a high-risk area, and the area with a score between A1 and A2 is a medium-risk area. Traverse the entire current grid map, calculate the score for each grid in the grid map, and divide the risk level according to the threshold; Adjacent grids of the same risk level are clustered to form continuous risk areas.

7. The indoor UAV flight path planning and processing method according to claim 6 is characterized in that: Based on multiple levels of partitioned areas, an improved rapidly expanding random tree algorithm is used to generate the initial flight path, including: Perform node sampling on the current multiple levels of partitioned areas. When sampling, assign different sampling frequencies to areas of different risk levels to reduce the sampling probability of high-risk areas; the sampling frequency gradually decreases from low-risk areas, medium-risk areas to high-risk areas; Set the starting point and the end point, run the improved RRT algorithm, and generate the initial flight path; when executing, the node at the starting point is used as the initial candidate node, and the distance between the current candidate node and multiple existing sampled tree nodes is calculated within the preset distance range; Within the preset distance range, select n tree nodes that are closer to the current candidate node as parent nodes, and use the local planner to connect the parent nodes and candidate nodes; Check whether the connection path intersects with obstacles; If they do not intersect, the candidate node is added to the tree; Repeat the above steps until a feasible path is found or the maximum number of iterations is reached and the generation of the initial flight path is completed.

8. The indoor UAV flight path planning and processing method according to claim 7 is characterized in that: After the initial flight path is generated, it also includes: The number of current initial flight paths is judged. If there are multiple current initial flight paths generated by the improved rapid expansion random tree algorithm, the number of areas of each risk level passed by the multiple current initial flight paths is calculated, and the initial flight path with the largest number of areas passing through the low-risk area is selected as the final target flight path.

9. The indoor UAV flight path planning and processing method according to claim 8, characterized in that: The process of smoothing the initial flight path includes: Use cubic spline interpolation to interpolate path points; Then the interpolated path points are subjected to a local optimization algorithm to adjust the path curvature; The performing of a local optimization algorithm to adjust the path curvature specifically includes: Calculate the curvature of each point on the path; For points where the curvature exceeds the threshold, the following optimization formula is applied: P'i=Pi+λ*SPi-1+Pi+1-2Pi+μ*SPi+2+Pi-2-2Pi); Among them, P'i is the optimized point coordinate, Pi is the current point coordinate, Pi-1, Pi+1, Pi-2 and Pi+2 are the adjacent point coordinates, λ and μ are weight coefficients; The above steps are iterated until the curvature meets the requirements or the maximum number of iterations is reached.

10. An indoor UAV flight path planning and processing system, which adopts the indoor UAV flight path planning and processing method according to any one of claims 1 to 8 for implementation, characterized in that: include: A three-dimensional map data acquisition module is used to acquire three-dimensional map data of an indoor environment; the three-dimensional map data is three-dimensional map data pre-constructed in a database; The laser radar scanning module is used to use the laser radar to scan the indoor environment and obtain real-time obstacle information; that is, the indoor environment is scanned by the laser radar device, so as to obtain multiple obstacle point cloud information and obstacle position information in the environment; A grid map construction module is used to construct a grid map based on three-dimensional map data and real-time obstacle information, obtain the density of obstacles and the probability value of occupied surfaces in each grid; divide the grid map into multiple levels of partition areas according to the density of obstacles and the probability value of occupied surfaces in the grid; A path generation module is used to generate an initial flight path based on multiple levels of partitioned areas using an improved rapidly expanding random tree algorithm; The path optimization module smoothes the initial flight path to obtain the optimized flight path; The flight control module is used to control the drone to fly according to the optimized flight path.

Images