An uninterrupted indoor and outdoor navigation mode selection method for a patrol robot constructed in a multi-sensor fusion architecture

Abstract

The application provides a kind of uninterrupted navigation mode selection method for indoor and outdoor of inspection robot constructed with multi-sensor fusion architecture, belongs to the field of robot navigation.The method comprises: obtaining the positioning information and environmental information of the inspection robot in real time through sensors and laser radars;Based on the obtained positioning information and environmental information, indoor and outdoor scene adaptive discrimination is carried out;According to the scene discrimination result, the path planning strategy is selected, the PRM algorithm is used for global path planning in outdoor scene, and the improved A* algorithm is used for global path planning in indoor scene;The inspection robot moves to the target position according to the planned path.The navigation mode of the application can flexibly adapt to different indoor and outdoor scenes, and meet the navigation requirements of the inspection robot in diversified application scenarios.

Description

Technical Field

[0001] This invention relates to a method for selecting indoor and outdoor uninterrupted navigation modes for inspection robots based on a multi-sensor fusion architecture, belonging to the field of robot navigation technology. Background Technology

[0002] With the increasing maturity of inspection robot technology, the application of robots to replace manual inspections is becoming more and more widespread. However, the widespread application of inspection robots highly depends on the efficiency and reliability of their navigation methods. Traditional inspection robot navigation technologies are often limited to single indoor or outdoor environments, making it difficult to cope with complex scenarios combining indoor and outdoor environments. In these complex scenarios, inspection robots need to traverse different environments efficiently and accurately, which places higher demands on navigation algorithms.

[0003] Existing navigation algorithms for inspection robots are mainly divided into search-based path planning algorithms and sampling-based path planning algorithms. Search-based path planning algorithms, such as breadth-first search and depth-first search, are inefficient in complex scenarios and prone to computational redundancy because they do not consider factors such as target location. While greedy best-first search algorithms improve path search efficiency, they are prone to getting trapped in local optima in complex environments. Dijkstra's algorithm, although guaranteeing the shortest path, is inefficient in complex scenarios. The A* algorithm combines the advantages of Dijkstra's and greedy best-first algorithms, but its path search efficiency still needs improvement in large-scale complex scenarios combining indoor and outdoor environments.

[0004] Path planning algorithms based on random sampling, such as PRM (Probabilistic Road Network) and RRT (Rapid Expanding Random Tree), perform well in specific environments, but perform poorly in complex obstacle distributions or narrow passable areas. Asymptotically optimal algorithms, such as RRT*, can approximate the optimal solution, but they have high computational costs and slow convergence speeds, making them difficult to meet the real-time requirements of navigation tasks.

[0005] Therefore, how to make full use of the various sensors equipped on the inspection robot to achieve efficient navigation control across scenarios has become a key factor in improving inspection efficiency and work results. Summary of the Invention

[0006] The purpose of this invention is to provide a method for selecting uninterrupted indoor and outdoor navigation modes for inspection robots using a multi-sensor fusion architecture. This addresses the problems of low navigation efficiency, suboptimal paths, and frequent detours, stops, and collisions in existing technologies for combined indoor and outdoor scenarios. Through multi-sensor fusion and adaptive scene recognition, efficient and safe automatic navigation of the inspection robot in combined indoor and outdoor environments is achieved.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] A method for selecting indoor and outdoor uninterrupted navigation modes for inspection robots based on a multi-sensor fusion architecture includes the following steps:

[0009] Step 1: Real-time acquisition of the inspection robot's positioning and environmental information through sensors and LiDAR. The positioning information includes linear velocity, angular velocity, and position change information, while the environmental information includes depth map and LiDAR point cloud information.

[0010] Step 2: Based on the acquired location and environmental information, perform adaptive indoor / outdoor scene discrimination, including:

[0011] The robot calculates the mean depth and depth difference along its forward direction to make an initial scene determination. If the scene satisfies a set of preset inequalities, it proceeds to the next step; otherwise, it is determined to be an indoor scene. The preset set of inequalities is as follows:

[0012] ,

[0013] in, The inspection robot's forward direction is close Frame depth mean The inspection robot's forward direction is close The maximum difference in pixel depth of a frame image. This represents the maximum measurable depth of the sensor. This represents the minimum measurable depth of the sensor.

[0014] If the scene is initially determined to be outdoor, then the depth map is divided into intervals and extracted. For each Harris 3D feature point, a plane is fitted using the RANSAC algorithm, and the maximum number of inliers is counted. If the maximum number of inliers is... If the following inequality is satisfied, proceed to the next step; otherwise, the scene is classified as indoors. The inequality is as follows: ;

[0015] The coordinate system plane of the inspection robot is defined with the front as the axis, and the left and right axes are... Each region is divided into two areas, and the number of point clouds whose observation distance on the left is the threshold of the maximum detection distance of the lidar is counted. The number of point clouds whose observation distance value on the right is the maximum detection distance threshold of the lidar. When satisfied If the condition is met, it is ultimately determined to be an outdoor scene; otherwise, it is determined to be an indoor scene. This refers to the scanning frequency of the lidar. This refers to the scanning resolution of the lidar.

[0016] Step 3: Select a path planning strategy based on the scene discrimination results. For outdoor scenes, the PRM algorithm combined with the improved A* algorithm is used for global path planning, while for indoor scenes, the improved A* algorithm is used for path planning.

[0017] Step 4: The inspection robot moves to the target location according to the planned path.

[0018] Preferably, the inspection robot's forward direction is close to... The frame depth mean is close to the direction of the inspection robot's movement. The maximum difference in pixel depth of a frame image is calculated as follows:

[0019] ,

[0020] in, The pixel width of the depth map. The pixel height of the depth map. This represents the depth information of each pixel in the depth map. This represents the maximum depth value that all pixels in the depth map can describe. This represents the minimum depth value that can be described by all pixels in the depth map. This indicates the linear velocity of the inspection robot. The sampling period represents the linear velocity and angular velocity. This indicates the angular velocity of the inspection robot. This indicates the x-coordinate of the inspection robot's position change. This represents the vertical coordinate of the inspection robot's position change. This indicates the sampling period for position change information.

[0021] Preferably, the inspection robot's sensors include internal sensors and external sensors; the internal sensors include an inertial unit and a wheel encoder, used to calculate linear velocity and angular velocity through extended Kalman filtering; the external sensors include an RGBD camera, used to acquire position transformation information and depth maps based on changes in landmark positions between adjacent frames.

[0022] Preferably, the improved A* algorithm includes:

[0023] The cost function is dynamically weighted and improved as follows:

[0024] ,

[0025] in, This represents the actual cost from the starting node to the current node n. This represents the estimated cost from the current node to the target node. This represents the Euclidean distance from the current node to the target node. This represents the Euclidean distance from the starting node to the target node.

[0026] When expanding a node, child nodes are filtered based on the direction of the target point, and nodes with an angle greater than 90 degrees to the main direction are closed.

[0027] Perform a second round of filtering on the remaining child nodes, and close the nodes located in the obstacles.

[0028] Preferably, the path planning of the PRM fusion improved A* algorithm specifically includes:

[0029] Density via PRM algorithm Sprinkle dots to eliminate obstructed areas;

[0030]

[0031] in, For extremely small values ​​greater than 0, map nodes are constructed using the removed points. After the nodes are constructed, all nodes are traversed and edges are constructed. Based on the edge construction, the edge path is optimized by improving the A* algorithm to generate smooth local paths.

[0032] Preferably, the method for filtering child nodes based on the target point direction is as follows:

[0033] During the node expansion process of the A* algorithm, the direction of the line connecting the current node and the target node is compared with the angles of the eight surrounding directions. The direction with the smallest angle is selected as the main direction, and potential child nodes in the three directions with an angle greater than 90 degrees from the main direction are closed.

[0034] Preferably, the secondary filtering of the remaining child nodes, closing nodes located in obstacles, is carried out in the following manner:

[0035] For the remaining child nodes after the initial screening, the child nodes located in the four directions of up, down, left, and right are screened a second time. If the child node is in an obstacle, the child node is closed and the expansion of subsequent child nodes is stopped.

[0036] Preferably, during the movement of the inspection robot, local dynamic obstacle detection is performed by fusing LiDAR with an RGBD camera, and the RGBD camera is used to detect obstacles within a depth range. To supplement the blind spots in the forward observation of the lidar;

[0037]

[0038] In the formula, It is the vertical distance between the center point of the rigid body of the lidar carried by the inspection robot and the ground. This refers to the elevation observation angle range of the lidar. This represents the maximum radial distance of the blind zone for nearby lidar observation.

[0039] The advantages of this invention are: through multi-sensor fusion and adaptive scene recognition, the inspection robot can efficiently search for optimal paths in indoor and outdoor combined scenarios, meeting the real-time navigation application needs in complex and large-scale scenarios.

[0040] By integrating RGBD cameras, the dynamic obstacle avoidance capability during navigation is enhanced, and by supplementing the detection blind spots of LiDAR, collisions are avoided during the inspection robot's navigation.

[0041] The improved A algorithm and PRM integration enhance the efficiency of path search and the quality of navigation paths, while reducing invalid inflection points and path extensions.

[0042] The navigation mode of this invention can flexibly adapt to different indoor and outdoor scenarios, meeting the navigation needs of inspection robots in diverse application scenarios. Attached Figure Description

[0043] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0044] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0045] Figure 2 This is a schematic diagram of the adaptive selection process of the navigation algorithm in an example of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example 1

[0048] like Figures 1-2 As shown, a method for selecting an uninterrupted indoor and outdoor navigation mode for an inspection robot using a multi-sensor fusion architecture is presented. The technical problem addressed is that when the inspection robot navigates in a combined indoor and outdoor environment, the significant structural differences between indoor and outdoor scenes make it difficult for conventional algorithms to simultaneously guarantee efficient path searching and optimal navigation path mileage. Furthermore, issues such as detours, pauses, and collisions frequently occur during navigation. The specific technical solution adopted is as follows:

[0049] Multi-sensor data fusion: The inspection robot obtains its own positioning information and environmental observation information through internal sensors (such as inertial units and wheel encoders) and external sensors (such as lidar and RGBD cameras).

[0050] Indoor and outdoor scene adaptive discrimination: Indoor and outdoor scenes are adaptively discriminated by fusing RGBD camera with LiDAR, and the current scene is judged by comprehensively using depth information, light intensity and geometric structure information.

[0051] Multi-strategy path planning: Different static global path planning algorithms are selected based on the scenario. For outdoor scenarios, the improved A algorithm (PRM fusion) is used for path planning, while for indoor scenarios, only the improved A algorithm is used. This enables uninterrupted indoor and outdoor navigation path planning for the inspection robot based on multi-sensor fusion.

[0052] Improved A* algorithm: The cost function of the original A* algorithm is designed with distance dynamic weighting to improve node search efficiency; during the node expansion process, child nodes are screened for the first and second time based on the target point location information, and child nodes with high risk coefficients are closed.

[0053] PRM is integrated with an improved A* algorithm: First, the PRM algorithm is used to perform coarse path search, generate a guiding path, and construct a sampling corridor area in the direction of the path; then, the PRM algorithm scatters points in the scene, removes obstacle points and constructs map nodes, and optimizes the path of each edge by using an improved A* algorithm.

[0054] Specifically, this embodiment includes the following steps:

[0055] S1: The robot's positioning and environmental information are acquired in real time through sensors and LiDAR. The positioning information includes linear velocity, angular velocity, and position change information. The environmental information includes depth map and LiDAR point cloud information.

[0056] As a refinement of the above embodiments, the sensors of the inspection robot include internal sensors and external sensors; the internal sensors include sensors with a sampling period of... The system includes an inertial unit and a wheel encoder for calculating linear and angular velocities using an extended Kalman filter; the external sensors include an RGBD camera for acquiring position change information based on changes in landmark positions between adjacent frames. And depth map; the sampling period of the RGBD device is The measurable depth range is .

[0057] S2: Based on the acquired location and environmental information, adaptive indoor and outdoor scene discrimination is performed, including:

[0058] S201: Calculate the mean depth and depth difference in the direction the inspection robot is moving to make a preliminary scene judgment. If the preset set of inequalities is satisfied, proceed to the next step; otherwise, determine it as an indoor scene. The preset set of inequalities is as follows:

[0059] ,

[0060] in, The inspection robot's forward direction is close Frame depth mean The inspection robot's forward direction is close The maximum difference in pixel depth of a frame image. For the maximum measurable depth, For the minimum measurable depth, This indicates rounding down, in this embodiment The value is 10.

[0061] As a refinement of the above embodiments, when all the inequalities in the system are satisfied, it is assumed that the inspection robot is currently in an outdoor scene based on scene depth information analysis.

[0062] The inspection robot is moving in a near direction. The frame depth mean is close to the direction of the inspection robot's movement. The maximum difference in pixel depth of a frame image is calculated as follows:

[0063] ,

[0064] in, The pixel width of the depth map. The pixel height of the depth map. This represents the depth information at each pixel location in the depth map. This represents the maximum depth value that all pixels in the depth map can describe. This represents the minimum depth value that can be described by all pixels in the depth map. This indicates the linear velocity of the inspection robot. The sampling period represents the linear velocity and angular velocity. This indicates the angular velocity of the inspection robot. This indicates the x-coordinate of the inspection robot's position change. This represents the vertical coordinate of the inspection robot's position change. This indicates the sampling period for position change information, i.e., the sampling period of the RGBD camera.

[0065] S202: If the scene is initially determined to be outdoor, then divide the depth map into intervals and extract the depth data. For each Harris 3D feature point, a plane is fitted using the RANSAC algorithm, and the maximum number of inliers is counted. If the maximum number of inliers is... If the following inequality is satisfied, proceed to the next step; otherwise, determine that it is an indoor scene.

[0066] As a refinement of the above embodiments, the depth map is first uniformly divided into intervals of pixel values ​​on the shorter side using a scene geometry analysis method. For each interval, the Harris 3D feature extraction algorithm is used to extract and retain one feature point, resulting in a total of [number missing] feature points extracted. We extract feature points, then use the RANSAC algorithm to fit a plane to obtain the maximum number of interior points N. Finally, we perform scene analysis and judgment. The inequalities are as follows:

[0067] ,

[0068] S203: Adjust the coordinate system plane of the inspection robot with the front as the axis, and move it left and right. Each region is divided into two areas, and the number of point clouds whose observation distance on the left is the threshold of the maximum detection distance of the lidar is counted. The number of point clouds whose observation distance value on the right is the maximum detection distance threshold of the lidar. When satisfied If the condition is met, it is ultimately determined to be an outdoor scene; otherwise, it is determined to be an indoor scene. This refers to the scanning frequency of the lidar. This refers to the scanning resolution of the lidar.

[0069] S3: Select a path planning strategy based on the scene discrimination results. For outdoor scenes, use the PRM algorithm combined with the improved A* algorithm for global path planning. For indoor scenes, use the improved A* algorithm for uninterrupted indoor and outdoor navigation path planning of the inspection robot based on multi-sensor fusion.

[0070] As a refinement of the above embodiments, the improved A* algorithm includes:

[0071] S301: Design a distance-dynamic weighted cost function for the original A* algorithm path search.

[0072] Specifically, the distance-dynamic weighted design of the cost function for path search in the original A* algorithm refers to dynamically weighting the actual cost from the initial position to the current position and the estimated cost from the current position to the target endpoint in order to improve the node search efficiency of the original A* algorithm. The cost function in the original A* algorithm is:

[0073] ,

[0074] In the above formula, This represents the actual cost from the starting node to the current node n. The improved cost function of the A* algorithm, which estimates the cost from the current node to the target node, is as follows:

[0075] ,

[0076] In the above formula, This represents the Euclidean distance from the current node to the target node. This represents the Euclidean distance from the starting node to the target node.

[0077] In the initial stage of navigation More emphasis on cost estimation This makes path search more conducive to rapidly expanding towards the target location. As the path search progresses, Dynamically improve The weighting of different factors is used to improve the quality of path search.

[0078] S302: During the node expansion process, the child nodes are initially screened based on the target point location information.

[0079] Specifically, the initial screening of child nodes based on the target point location information refers to comparing the direction of the line connecting the current node and the target node with the angles of the eight surrounding directions during the node expansion process of the A* algorithm, selecting the direction with the smallest angle as the main direction, and closing potential child nodes in the three directions with an angle greater than 90 degrees from the main direction.

[0080] S303: After the initial screening of child nodes, a second screening is performed and child nodes with high risk factors are closed.

[0081] Specifically, performing a second screening and closing high-risk child nodes after the initial screening means that among the five remaining child nodes after the initial screening, the child nodes located in the four directions of up, down, left, and right are screened again. If the child node is in an obstacle, the child node is closed and the subsequent expansion of child nodes is stopped.

[0082] As a refinement of the above embodiments, the path planning of the PRM fusion improved A* algorithm specifically includes:

[0083] Density via PRM algorithm Sprinkle dots to eliminate obstructed areas;

[0084] ,

[0085] in, To increase the density, To improve the A* algorithm for estimating the cost function, The value is a very small number greater than 0. If the grid map information of the point's location is 1, it is considered an obstacle (the grid map is composed of countless small grids, each with a value of 0 and 1. 0 represents empty space, and the visual color is white; 1 represents occupied space, and the visual color is black. Based on the location information of the scattered points, we can determine whether there is an obstacle by checking whether the corresponding grid information is 1 or 0). After removing the points on the obstacles, the remaining points are used to construct the nodes of the map. After the nodes are constructed, all nodes are traversed and edges are constructed. Then, the edges are used to guide the improvement of the A* algorithm to optimize the path of each edge and generate a smooth local path.

[0086] It's important to note that when the inspection robot determines it's switching between indoor and outdoor environments during navigation, it performs a static global path search again. For example, if it starts outdoors and activates the PRM fusion improved A* algorithm, but then determines it's indoors, it switches to the improved A* algorithm from the current location to the destination. The reverse applies for indoor-outdoor transitions. Indoor environments generally have more obstacles, while outdoor navigation involves longer distances. The advantage of PRM over A* is that, when there are fewer obstacles and the distance is greater, its search efficiency and computational cost are significantly higher. The disadvantage is that in environments with many obstacles and limited space, the searched path is often not optimal. In larger scenarios, a local incremental static global path planning strategy is used, such as the A* algorithm. This involves searching a segment of static global path first, and then extending it to a new segment after a certain distance. It doesn't reach the target path in one step because the search time is long and the computational cost increases exponentially, consuming significant processor resources. The improved A* algorithm used in this patent is also a local incremental method; it won't use A* to search for a path spanning both the starting and ending points indoors (if the starting and ending points exceed the search segment distance threshold).

[0087] S4: The inspection robot moves to the target location according to the planned path.

[0088] As a refinement of the above embodiments, the inspection robot performs local dynamic obstacle detection by fusing LiDAR and RGBD cameras during its movement, and the fusion depth range is within... The RGBD observation information supplements the environmental information of the lidar, thus avoiding collisions when the inspection robot navigates to indoor scenes such as long corridors and structural corners.

[0089] ,

[0090] In the formula, It is the vertical distance between the center point of the rigid body of the lidar carried by the inspection robot and the ground. This refers to the elevation observation angle range of the lidar. This represents the maximum radial distance of the blind zone for nearby lidar observation.

[0091] When the inspection robot performs path planning, due to its hardware characteristics, the LiDAR inevitably has a point in the XZ plane with height h and an acute angle of θ. The right-angled triangle detection blind zone can be effectively prevented from colliding with dynamic obstacles when the robot moves or turns in place at indoor corners by fusing depth observation information from the RGBD camera in this area to supplement environmental observation.

[0092] It should be noted that this invention leverages the complementary advantages of multiple sensors commonly carried by inspection robots, such as LiDAR and RGBD cameras. Furthermore, considering the structural and functional differences between indoor and outdoor scenarios, a strategy for adaptive scene recognition and selection of different path methods is specifically designed. In addition, to improve path search efficiency and navigation safety, this invention incorporates an improved A* algorithm and RGBD dynamic obstacle observation constraints, thereby meeting the requirements for efficient and safe automatic navigation of inspection robots in combined indoor and outdoor scenarios.

[0093] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture, characterized in that, Includes the following steps: Step 1: Acquire the inspection robot's positioning information and environmental information in real time through sensors and LiDAR. The positioning information includes linear velocity, angular velocity, and position change information. The environmental information includes depth map and LiDAR point cloud information. Step 2: Based on the acquired location and environmental information, perform adaptive indoor / outdoor scene discrimination, including: The robot calculates the mean depth and depth difference along its forward direction to make an initial scene determination. If the scene satisfies a set of preset inequalities, it proceeds to the next step; otherwise, it is determined to be an indoor scene. The preset set of inequalities is as follows: ， in, The inspection robot's forward direction is close Frame depth mean The inspection robot's forward direction is close The maximum difference in pixel depth of a frame image. This represents the maximum measurable depth of the sensor. This represents the minimum measurable depth of the sensor. If the scene is initially determined to be outdoor, then the depth map is divided into intervals and extracted. For each Harris 3D feature point, a plane is fitted using the RANSAC algorithm, and the maximum number of inliers is counted. If the maximum number of inliers is... If the following inequality is satisfied, proceed to the next step; otherwise, the scene is classified as indoors. The inequality is as follows: ; The coordinate system plane of the inspection robot is defined with the front as the axis, and the left and right axes are... Each region is divided into two areas, and the number of point clouds whose observation distance on the left is the threshold of the maximum detection distance of the lidar is counted. The number of point clouds whose observation distance value on the right is the maximum detection distance threshold of the lidar. When satisfied If the condition is met, it is ultimately determined to be an outdoor scene; otherwise, it is determined to be an indoor scene. This refers to the scanning frequency of the lidar. This refers to the scanning resolution of the lidar. Step 3: Select a path planning strategy based on the scene discrimination results. For outdoor scenes, use the PRM fusion improved A* algorithm for global path planning, and for indoor scenes, use the improved A* algorithm for path planning. Step 4: The inspection robot moves to the target location according to the planned path.

2. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 1, characterized in that, The inspection robot is moving in a near direction. The frame depth mean is close to the direction of the inspection robot's movement. The maximum difference in pixel depth of a frame image is calculated as follows: ， in, The pixel width of the depth map. The pixel height of the depth map. This represents the depth information of each pixel in the depth map. This represents the maximum depth value that all pixels in the depth map can describe. This represents the minimum depth value that can be described by all pixels in the depth map. This indicates the linear velocity of the inspection robot. The sampling period represents the linear velocity and angular velocity. This indicates the angular velocity of the inspection robot. This indicates the horizontal coordinate of the inspection robot's position change. This represents the vertical coordinate of the inspection robot's position change. This indicates the sampling period for position change information.

3. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 1, characterized in that, The inspection robot's sensors include internal and external sensors; the internal sensors include an inertial unit and a wheel encoder, used to calculate linear velocity and angular velocity through extended Kalman filtering; the external sensors include an RGBD camera, used to acquire position transformation information and depth maps based on changes in landmark positions between adjacent frames.

4. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 1, characterized in that, The improved A* algorithm includes: The cost function is dynamically weighted and improved as follows: ， in, This represents the actual cost from the starting node to the current node n. This represents the estimated cost from the current node to the target node. This represents the Euclidean distance from the current node to the target node. This represents the Euclidean distance from the starting node to the target node. When expanding a node, child nodes are filtered based on the direction of the target point, and nodes with an angle greater than 90 degrees to the main direction are closed. Perform a second round of filtering on the remaining child nodes, and close the nodes located in the obstacles.

5. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 4, characterized in that, The path planning of the PRM fusion improved A* algorithm specifically includes: Density via PRM algorithm Sprinkle dots to eliminate obstructed areas; in, For extremely small values ​​greater than 0, map nodes are constructed using the removed points. After the nodes are constructed, all nodes are traversed and edges are constructed. Based on the constructed edges, the edge paths are optimized by improving the A* algorithm to generate smooth local paths.

6. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 4, characterized in that, The specific method for filtering child nodes based on the target point direction is as follows: During the node expansion process of the A* algorithm, the direction of the line connecting the current node and the target node is compared with the angles of the eight surrounding directions. The direction with the smallest angle is selected as the main direction, and potential child nodes in the three directions with an angle greater than 90 degrees from the main direction are closed.

7. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 6, characterized in that, The second filtering of the remaining child nodes, closing nodes located in obstacles, is performed as follows: For the remaining child nodes after the initial screening, the child nodes located in the four directions of up, down, left, and right are screened a second time. If the child node is in an obstacle, the child node is closed and the expansion of subsequent child nodes is stopped.

8. The method for selecting indoor and outdoor uninterrupted navigation modes for an inspection robot based on a multi-sensor fusion architecture as described in claim 1, characterized in that, The inspection robot performs local dynamic obstacle detection by fusing lidar and RGBD cameras during its movement, and uses the RGBD cameras to detect obstacles within a depth range. To supplement the blind spots in the forward observation of the lidar; In the formula, It is the vertical distance between the center point of the rigid body of the lidar carried by the inspection robot and the ground. This refers to the elevation observation angle range of the lidar. This represents the maximum radial distance of the blind zone for nearby lidar observation.

Images