Autonomous exploration method and system of mobile robot in unknown environment

Abstract

The invention discloses an autonomous exploration method and system for a mobile robot in an unknown environment, and the method comprises the steps: firstly scanning the environment in real time to obtain the positioning and point location information of the robot, and updating a 3D grid map represented by a three-dimensional space; secondly, boundary detection and candidate viewpoint sampling are carried out on a map, and viewpoint weights are calculated to screen extension nodes; carrying out path planning by adopting a Hybrid A * algorithm, and maintaining a path into a topological map; path tracking is carried out by adopting a pure path tracking method based on a ray distance, the speed is adjusted in combination with a PID algorithm, and the accuracy and real-time performance of path tracking are improved. According to the method, the completeness of expanded viewpoint selection is improved, repeated exploration of the same area is reduced, the path tracking precision is improved, and the technical problem of autonomous exploration of the robot in a large-scale unknown environment is effectively solved.

Description

Technical Field

[0001] The present invention relates to a method and system for autonomous exploration of a robot, specifically a method and system for autonomous exploration of a mobile robot in an unknown environment, belonging to the technical field of mobile robots. Background Art

[0002] With the development of automation technology, ground robots are increasingly widely used in various industrial and exploration fields. Especially in underground working environments such as coal mines, robots can perform dangerous or difficult-to-reach tasks for humans. However, existing ground robot autonomous navigation systems usually rely on pre-set maps and environmental information, which is impractical in unknown or dynamically changing scenarios. Therefore, it has become particularly important to develop an autonomous navigation method for mobile robots that can effectively explore autonomously in large-scale unknown scenarios.

[0003] However, fully and efficiently exploring a large-scale environment remains a challenging open problem. Especially in the absence of a prior map, the navigation system faces great uncertainties in the path planning and decision-making processes. Such uncertainties not only increase the complexity of the task but may also lead to navigation failures or low efficiency. Therefore, how to perform effective path planning and decision-making in unknown or partially known environments has become the focus of research in recent years.

[0004] Traditional path planning methods usually rely on optimization techniques to generate a feasible path from the starting point to the target point by minimizing a specific objective function (such as path length, energy consumption, or time). These methods gradually build an environmental map during the exploration process and use the incremental map information for real-time planning. However, this method has obvious limitations: First, incremental map construction and maintenance require a large amount of computing resources, especially in large-scale environments, where the computational overhead will increase significantly; Second, the process of constructing and updating the global map may lead to planning delays, affecting the real-time performance of the system.

[0005] In terms of decision-making, many methods perceive the environment by analyzing the geometric features of the environment (such as the shape, distance, and distribution of obstacles) and make optimal decisions based on these features. For example, some algorithms plan paths by detecting key points or boundaries in the environment or use local information for obstacle avoidance. However, these methods usually rely on the maintenance of the global map, resulting in low computational efficiency. In addition, due to the uncertainty and dynamic changes of the environment, decision-making methods based on geometric features may not be able to adapt to complex or unstructured terrains. Especially in the case of lateral error accumulation, the navigation accuracy will decrease significantly.

[0006] Conventional path tracking methods (such as the pure pursuit algorithm) perform well in structured environments but often struggle to adapt in unstructured terrains. The pure pursuit algorithm guides the robot's movement by tracking a predefined path. However, when faced with complex terrains, lateral errors gradually accumulate, leading to path deviation and even navigation failure. This limitation is particularly prominent in environments lacking prior maps because the robot cannot anticipate terrain changes in advance and is thus difficult to make timely adjustments.

[0007] In summary, although a large amount of research has been dedicated to path planning and decision-making in unknown or partially unknown environments, existing methods still have significant deficiencies in terms of computational efficiency, real-time performance, and adaptability. Especially in large-scale, unstructured environments, how to achieve efficient and accurate exploration without relying on a global map remains an urgent problem to be solved. Summary of the Invention

[0008] The objective of the present invention is to provide a method for autonomous exploration of a mobile robot in an unknown environment, which can improve the completeness of the selection of extended viewpoints, reduce repeated exploration of the same area, and improve path tracking accuracy, effectively solving the technical problems of autonomous exploration of robots in large-scale unknown environments, and can be applied to the autonomous exploration of mobile robots in large-scale unknown scenarios.

[0009] To achieve the above objective, the present invention provides a method for autonomous exploration of a mobile robot in an unknown environment, including the following steps:

[0010] Step S1: Obtain the real-time positioning information and point cloud information of the robot, register the point cloud information, and update the 3D grid map representing the three-dimensional space using the point cloud data information;

[0011] Step S2: Perform incremental boundary detection in the updated 3D grid map, sample candidate viewpoints, and evaluate the sampled viewpoints to screen out the nodes to be expanded;

[0012] Step S3: In the area where the robot detects that it can move freely, use the search-based HybridA* path planning algorithm to search for paths from the current robot position to all candidate viewpoints obtained by sampling, and maintain the paths in the topological map;

[0013] Step S4: Use the pure path tracking method based on dynamic ray distance to track the path from the current robot position to the expanded node.

[0014] The autonomous exploration method of a mobile robot in an unknown environment according to an embodiment of the present invention first obtains the real-time positioning information and point cloud information of the robot, and converts the information into a grid map format to construct or update a 3D grid map; then, performs boundary detection on the updated 3D grid map, samples candidate viewpoints that meet the requirements, and evaluates all viewpoints to screen out target nodes (i.e., expansion nodes); then, uses the search-based HybridA* path planning algorithm to search within the area where the robot can move freely, and maintains the searched path in the topological map, where the searched path includes the paths from the current robot position to all candidate viewpoints; and then uses a pure path tracking method based on dynamic ray distance to track the path from the current robot position to the target node to achieve path tracking. It should be noted that during the tracking process, the real-time positioning information and point cloud information of the robot will change, and the 3D grid map will be updated in real time. Therefore, after tracking to the target node, the robot will enter the exploration and tracking of the next cycle, and so on, to achieve autonomous exploration in a large-scale unknown environment.

[0015] In some embodiments of the present invention, step S1 specifically includes:

[0016] Step S11: Use a lidar-based SLAM module to scan the environment, obtain the positioning and point cloud information of the robot, process the point cloud information, and convert it into a grid map representation for real-time updating of the 3D grid map;

[0017] Step S12: Each node in the 3D grid map corresponds to a cubic region in the environmental space. When new point cloud data arrives, the 3D grid map updates the state of the corresponding cubic region; among them, in the initial state, all regions are marked as unknown space S uk ; as the robot moves and the SLAM module scans, the unknown regions will be gradually explored and updated to the known state, and the known state is divided into known free space S free and known occupied space S occ , and the known free space S free is the area where the robot can move freely detected by the sensor; the known occupied space S occ is the impassable area detected by the robot through the sensor.

[0018] In some embodiments of the present invention, step S2 specifically includes:

[0019] Step S21: Perform boundary detection according to the updated 3D grid map, and only detect the newly added grids recorded in the 3D grid map and the existing boundaries of the robot;

[0020] Step S22, uniform random sampling is performed around the frontier boundary detected in the area where the robot can move freely through the sensor to obtain candidate viewpoints;

[0021] Step S23: Perform multi-dimensional viewpoint comprehensive evaluation on the randomly sampled candidate viewpoints to screen out suitable candidate viewpoints.

[0022] In some embodiments of the present invention, the boundary in step S21 is: in the known 3D space, if there is a free grid with at least one unknown neighbor grid of the same height, then the free grid is defined as the boundary; let p i is the i-th position of the robot, For robots in p i The global boundary at For p i to p i+1 The boundaries within the robot's range, For robots from p i Move to p i+1 The updated sensor senses a free grid within the range, then Boundary detection is performed in .

[0023] In some embodiments of the present invention, the extended nodes screened in step S23 are suitable candidate viewpoints screened out by multi-dimensional viewpoint comprehensive weight calculation, and the weights include information gain, travel cost and topological features, as follows:

[0024] The information gain is evaluated by comprehensively considering the number of unknown grids visible from the viewpoint. The calculation is performed by simulating the ray casting method of the lidar. A simulated laser ray is emitted from the viewpoint to the detected frontier boundary. The formula Calculate the number of grids covering the unknown area, where δ i represents the contribution value of the i-th unknown grid, and P(v) represents the information gain of viewpoint v;

[0025] The travel cost is based on the topological structure of the map, including the actual shortest path length L(v0,v) from the current node to the candidate viewpoint and the number of nodes L passed by the path. t (v0,v), where the shortest path length is searched using the HybridA* algorithm;

[0026] The topological feature explores the endpoints of the topological map based on the position of the viewpoint in the topological map, where the endpoints of the topological map represent a complete area to be explored, and the nodes are the intersections of different areas;

[0027] Therefore, the viewpoint weight calculation formula is: Among them, U(v k ) is the viewpoint vk The utility of G(v k ) is the penalty function, P(v k ) is the information gain, c is the penalty factor, and L(v0→v k ) is the moving distance from the current viewpoint to viewpoint v k .

[0028] In some embodiments of the present invention, the pure path tracking method based on the dynamic ray distance in S4 tracks a predetermined path by controlling the linear velocity and steering angle of the robot, and the specific steps are as follows:

[0029] Step S41: Calculate the ray distance and the required steering angle according to the current position of the robot, the target position, and the next target point on the path;

[0030] Step S42: Calculate the ray distance, which refers to the straight-line distance from the current position of the robot to the target point. The ray distance is dynamically adjusted according to the speed of the robot and the minimum turning radius, and the calculation formula is: l d = k*V + C, where k is the speed coefficient, V is the speed of the robot, and C is the minimum turning radius of the robot;

[0031] Step S43: Calculate the steering angle. The steering angle θ is calculated through the Ackermann steering model and depends on the wheelbase L and the turning radius R of the vehicle. The calculation formula is: tanθ = L / R;

[0032] Step S44: Use the PID algorithm to adjust the speed of the robot to ensure that the robot travels along the predetermined path.

[0033] In some embodiments of the present invention, during the path tracking process of step S44, the position error of the robot is used to adjust the steering angle to reduce the movement deviation of the robot and keep it on the path; among them, the position error is the lateral error e(t).

[0034] The present invention also discloses a mobile robot autonomous exploration system in an unknown environment, including:

[0035] The SLAM module is used to receive the positioning information of the robot and the real-time point cloud information around the vehicle body collected by the sensors on the robot body, register the point cloud information to perform point cloud data modeling, and construct and update the 3D grid map;

[0036] The path analysis module is used to receive the 3D grid map information transmitted by the SLAM module, perform map frontier detection, generate candidate viewpoints based on sampling at the map frontier, and screen out expansion nodes through a multi-dimensional viewpoint comprehensive evaluation mechanism;

[0037] The path planning module generates paths from the current position of the robot to all candidate viewpoints through the search-based HybridA* path planning algorithm, and maintains the paths in the topological map.

[0038] The path tracking module. The robot selects the expansion nodes based on the path analysis module, obtains the paths passing through the expansion nodes through the path planning module, and uses the pure path tracking method based on ray distance to track the paths. Description of the Drawings

[0039] Figure 1 is a flowchart of a mobile robot autonomous exploration method in an unknown environment according to an embodiment of the present invention;

[0040] Figure 2 is a schematic diagram of the system of the present invention;

[0041] Figure 3 is a schematic diagram of the overall process of autonomous exploration of the present invention;

[0042] Figure 4 is a schematic diagram of the process of detecting the boundary to find the target nodes in the present invention;

[0043] Figure 5 is a schematic diagram of the principle of the pure path tracking method based on dynamic ray distance in the present invention. Detailed Embodiment

[0044] The embodiments of the present invention are described in detail below. The examples of the embodiments are shown in the drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0045] The following references Figures 1 - 5 describe a mobile robot autonomous exploration method in an unknown environment according to an embodiment of the present invention, and the method includes:

[0046] Step S1: Obtain the real-time positioning information and point cloud information of the robot, register the point cloud information, and update the 3D grid map representing the three-dimensional space using the point cloud data information;

[0047] Step S2: Perform incremental boundary detection in the updated 3D grid map, sample candidate viewpoints, and evaluate the sampled viewpoints to select the nodes to be expanded;

[0048] Step S3: In the area where the robot detects that it can move freely, use the search-based HybridA* path planning algorithm to search for paths from the current position of the robot to all candidate viewpoints obtained by sampling, and maintain the paths in the topological map;

[0049] Step S4: Use the pure path tracking method based on dynamic ray distance to track the path from the current position of the robot to the expansion node.

[0050] It can be understood that in the actual exploration and tracking process, first obtain the real-time positioning information and point cloud information of the robot, and convert the information into a grid map format to construct or update the 3D grid map; then, perform boundary detection on the updated 3D grid map, sample candidate viewpoints that meet the requirements, and evaluate all viewpoints to screen out the target node (i.e., the expansion node); then, use the search-based HybridA* path planning algorithm to search in the area where the robot can move freely, and maintain the searched path in the topological map, where the searched path includes the paths from the current robot position to all candidate viewpoints; then use the pure path tracking method based on dynamic ray distance to track the path from the current position of the robot to the target node to achieve path tracking. It should be noted that during the tracking process, the real-time positioning information and point cloud information of the robot will change, and the 3D grid map will be updated in real time. Therefore, after tracking to the target node, the robot will enter the exploration and tracking of the next cycle, and so on, to achieve autonomous exploration in a large-scale unknown environment.

[0051] In view of this, the movement of the robot is divided into path search and path tracking. The path is obtained by searching with the search-based path planning algorithm HybridA*. Path tracking uses a pure path tracking method based on ray distance proposed in the present invention, which is a path tracking technology for mobile robots. It is particularly suitable for car-like mobile robots with an Ackermann steering model, aiming to improve the accuracy and real-time performance of path tracking, reduce tracking errors, and at the same time meet the dynamic constraints of the mobile robot.

[0052] In some embodiments of the present invention, step S1 specifically includes:

[0053] Step S11: Use the SLAM module based on lidar to scan the environment, obtain the positioning and point cloud information of the robot, and process the point cloud information to convert it into a grid map representation for real-time updating of the 3D grid map; among them, the 3D grid map used is OctoMap, which is used to represent the three-dimensional environment. Its nodes store the metrics of the space represented by its child nodes, allowing some branches to be ignored when traversing the octree, and each node corresponds to a Morton code for quickly querying the node, thereby accelerating the traversal.

[0054] Step S12. Each node in the 3D grid map corresponds to a cubic region in the environmental space. When new point cloud data arrives, the 3D grid map updates the status of the corresponding cubic region. Specifically, each node in OctoMap corresponds to a cubic region (voxel) in the environmental space. When new point cloud data arrives, OctoMap updates the status of the corresponding voxel, which specifically includes the following steps:

[0055] In the initial state, all regions are marked as unknown space S uk ; As the robot moves and the SLAM module scans, the unknown regions will be gradually explored and updated to the known state. The known state is divided into known free space S free and known occupied space S occ ; The known free space S free is the region where the robot can move freely detected by the sensor; The known occupied space S occ is the impassable region detected by the robot through the sensor.

[0056] In some embodiments of the present invention, step S2 specifically includes:

[0057] Step S21. Perform boundary detection based on the updated 3D grid map. To improve the boundary detection efficiency and avoid searching the entire map every time the map is updated, only detect the newly added grids recorded in the 3D grid map and the existing boundaries of the robot;

[0058] Step S22. Perform uniform random sampling around the detected frontiers in the region where the robot can move freely detected by the sensor (i.e., in the known free space S free ) to obtain candidate viewpoints;

[0059] Step S23. Perform multi-dimensional viewpoint comprehensive evaluation on the candidate viewpoints obtained by random sampling to select suitable candidate viewpoints.

[0060] In some embodiments of the present invention, the boundary in step S21 is: in the known 3D space, if there is a free grid with at least one unknown neighbor grid of the same height, then the free grid is defined as the boundary; Let p i be the i-th position of the robot, be the global boundary of the robot at p i , be the boundary within the range of the robot from p i to p i+1 , be the free grid within the updated sensor perception range when the robot moves from p i to p i+1 , then Because boundaries must appear in the updated grid, some existing boundaries may no longer be boundaries due to the neighboring grids being updated.

[0061] In some embodiments of the present invention, the extended nodes screened in step S23 are suitable candidate viewpoints screened out by multi-dimensional viewpoint comprehensive weight calculation, and the weights include information gain, travel cost and topological features, as follows:

[0062] The information gain is evaluated by comprehensively considering the number of unknown grids visible from the viewpoint. The calculation is performed by simulating the ray casting method of the lidar. A simulated laser ray is emitted from the viewpoint to the detected frontier boundary. The formula Calculate the number of grids covering the unknown area, where δ i represents the contribution value of the i-th unknown grid, and P(v) represents the information gain of viewpoint v;

[0063] The travel cost is based on the topological structure of the map, including the actual shortest path length L(v0,v) from the current node to the candidate viewpoint and the number of nodes L passed by the path. t (v0,v), where the shortest path length is searched using the HybridA* algorithm;

[0064] The topological feature explores the endpoints of the topological map based on the position of the viewpoint in the topological map. The endpoints of the topological map represent a complete area to be explored, and the nodes are the intersections of different areas. Prioritizing the exploration of the endpoints of the topological map can ensure that different areas are explored in sequence, avoiding unnecessary repeated exploration.

[0065] Therefore, the multi-dimensional viewpoint weight calculation formula is: Among them, U(v k ) is the viewpoint v k The utility of G(v k ) is the penalty function, P(v k ) is the information gain, c is the penalty factor, L(v0→v k ) is the distance from the current viewpoint to the viewpoint v k moving distance.

[0066] At the candidate viewpoint v k In step S2, the viewpoint weight calculated in step S2 is used to select the candidate viewpoint with the largest weight value as the extended node v'0, so that the robot moves from the current node to the extended node (target node).

[0067] It should be noted that the selection of extended nodes by adopting the multi-dimensional view point comprehensive evaluation mechanism fully considers the information gain, travel cost, and topological features of candidate nodes. Compared with the prior art that only considers information gain, it improves the completeness of the selection of extended view points and effectively avoids the deficiency of the prior art that is prone to falling into local optimality when performing view point selection.

[0068] In some embodiments of the present invention, in S4, the pure path tracking method based on the dynamic ray distance controls the linear velocity and steering angle of the robot to track a predetermined path, and the specific steps are as follows:

[0069] Step S41: Calculate the ray distance and the required steering angle according to the current position of the robot, the target position, and the next target point on the path (i.e., the extended node, the same below);

[0070] Step S42: Calculate the ray distance. The ray distance refers to the straight-line distance from the current position of the robot to the target point, and the ray distance is dynamically adjusted according to the speed of the robot and the minimum turning radius, which is used to determine the forward-looking when tracking the path; its calculation formula is: l d = k*V + C, where k is the speed coefficient, V is the speed of the robot, and C is the minimum turning radius of the robot;

[0071] Step S43: Calculate the steering angle. The steering angle θ is calculated through the Ackermann steering model and depends on the wheelbase L and the turning radius R of the vehicle. The calculation formula is: tanθ = L / R;

[0072] Step S44: Use the PID algorithm to adjust the speed of the robot to ensure that the robot travels along the predetermined path.

[0073] In some embodiments of the present invention, during the path tracking process of step S44, the position error of the robot is used to adjust the steering angle to reduce the movement deviation of the robot and keep it on the path; wherein, the position error is the lateral error e(t).

[0074] The present invention also discloses a mobile robot autonomous exploration system in an unknown environment, including:

[0075] The SLAM module is used to receive the positioning information of the robot and the real-time point cloud information around the vehicle body collected by the sensors on the robot body, register the point cloud information to perform point cloud data modeling, and construct and update the 3D grid map;

[0076] The path analysis module is used to receive the 3D grid map information transmitted by the SLAM module, perform map front detection, generate candidate view points based on sampling at the map front, and screen out extended nodes through the multi-dimensional view point comprehensive evaluation mechanism;

[0077] The path planning module generates paths from the current position of the robot to all candidate viewpoints through the search-based HybridA* path planning algorithm and maintains the paths in the topological map.

[0078] The path tracking module. The robot filters out expansion nodes based on the path analysis module, obtains the paths passing through the expansion nodes through the path planning module, and uses the pure path tracking method based on ray distance to track the paths.

[0079] It should be noted that the system of the present invention includes a SLAM module, a path analysis module, a path planning module, and a path tracking module; the information output by the SLAM module is transmitted to the path analysis module, the information output by the path analysis module is transmitted to the path planning module, the information output by the path planning module is transmitted to the path tracking module, and finally the path tracking module is connected to the SLAM module for further exploration and tracking.

[0080] Specifically, the functions and roles of each module are as follows:

[0081] I. SLAM module:

[0082] In this embodiment, the SLAM module, as the upstream module of path planning and control, scans the surrounding environment of the robot in real time, obtains the positioning information of the fuselage positioning and perception and the point cloud data information after registration, and uses the point cloud data information to update the OctoMap in real time.

[0083] Each node in the OctoMap corresponds to a cubic region (voxel) in the environmental space. When new point cloud data arrives, the OctoMap updates the status of the corresponding voxel, specifically including:

[0084] In the initial state, all regions are marked as unknown space S uk ;

[0085] As the robot moves and the SLAM module scans, the unknown regions will be gradually explored and updated to the known state, and the known state is divided into the following two parts:

[0086] Known free space S free : The area that the robot can move freely through the sensor detection.

[0087] Known occupied space S occ : The non-passable area that the robot detects through the sensor.

[0088] II. Path analysis module:

[0089] In this implementation example, the path analysis module processes the map information provided by the front-end SLAM module. This module is a key module for the robot to complete autonomous exploration tasks in unknown environments. Appropriate candidate viewpoint sampling and evaluation methods can greatly improve the efficiency of the robot's autonomous exploration tasks.

[0090] like Figure 4 As shown, in this implementation example, the path analysis module specifically includes the following steps:

[0091] First, a boundary detection is performed quickly in the updated 3D grid map, where a boundary is defined as a free grid with at least one unknown neighbor grid of the same height in the known 3D space. To improve the efficiency of boundary detection, only the newly added grids recorded in the 3D grid map and the existing boundaries of the mobile robot need to be detected.

[0092] Let p i is the i-th position of the robot, For robots in p i The global boundary at For p i to p i+1 The boundaries within the robot's range, For robots from p i Move to p i+1 The updated sensor sensing range is free grid; then Because boundaries must appear in the updated grid, some existing boundaries may no longer be boundaries due to the update of neighboring grids.

[0093] Secondly, in the known free space S free Uniform random sampling is performed around the frontier boundary detected in to obtain candidate viewpoints.

[0094] Next, the viewpoint comprehensive weight is calculated for the candidate viewpoints obtained by random sampling, where the weight comprehensively considers information gain, travel cost, and topological characteristics, specifically including the following steps:

[0095] The information gain is evaluated, which comprehensively considers the number of unknown grids visible from the viewpoint and is calculated by simulating the ray casting method of LiDAR.

[0096] A simulated laser ray is emitted from the viewpoint to the detected frontier boundary, and the formula Calculate the number of grids covering the unknown area, where δ i represents the contribution value of the i-th unknown grid, and P(v) represents the information gain of viewpoint v.

[0097] The travel cost is based on the topology of the map, including the actual shortest path length L(v0, v) from the current node to the candidate viewpoints and the number of nodes L(v0, v) passed by this path, where the shortest path length is obtained by searching using the HybridA* algorithm. t (v0, v), where the shortest path length is obtained by searching using the HybridA* algorithm.

[0098] The topological features are considered based on the position of the viewpoints in the topological graph. The endpoints of the topological graph usually represent a complete area to be explored, while the nodes are the intersection points of different areas. Exploring the endpoints of the topological graph first can ensure that different areas are explored in sequence and avoid unnecessary repeated exploration.

[0099] Taking all these into consideration, the viewpoint weight is calculated as: where U(v k ) is the utility of the viewpoint v k , G(v k ) is the penalty function, P(v k ) is the information gain, c is the penalty factor, L(v0→v k ) is the moving distance from the current viewpoint to the viewpoint v k .

[0100] III. Path Planning Module:

[0101] In this embodiment, the path planning module uses the search-based HybridA* path planning algorithm to search for paths from the current robot position to all candidate viewpoints. Among them, the path that the robot moves from the current position to the optimal viewpoint (i.e., the expanded node) screened by the path analysis module is searched and handed over to the path tracking module for tracking, and the remaining paths are maintained in the topological graph.

[0102] IV. Path Tracking Module:

[0103] In this embodiment, the path tracking module tracks the path from the current robot position to the expanded node transmitted by the path planning module, and based on the requirements of path smoothness and tracking accuracy, uses a pure path tracking method based on ray distance to track the path.

[0104] In this embodiment, the pure path tracking method based on ray distance tracks the predetermined path by controlling the linear velocity and steering angle of the robot. During the path tracking process, the current position, target position of the robot and the next target point on the path (i.e., the expanded node) are used to calculate the ray distance and the required steering angle. The calculation methods of the ray distance and the steering angle are as follows:

[0105] The Line-of-Sight Distance refers to the straight-line distance from the current position of the robot to the target point, which can be dynamically adjusted according to the speed of the robot and the minimum turning radius, and is used to determine the look-ahead when tracking the path. The calculation formula for the Line-of-Sight Distance is: l d = k * V + C, where k is the speed coefficient, V is the speed of the robot, and C is the minimum turning radius of the robot.

[0106] The steering angle θ is calculated through the Ackermann steering model, and its calculation formula is: tanθ = L / R, where L is the wheelbase of the vehicle and R is the turning radius.

[0107] The path tracking control uses the PID (Proportional-Integral-Derivative) algorithm to adjust the speed of the robot to ensure that the robot travels along the predetermined path. During the path tracking process, the position error (lateral error e(t)) of the mobile robot is used to adjust the steering angle to reduce the deviation and stay on the path, making the robot track the path more smoothly and accurately.

[0108] In summary, in view of the problem of the relatively high difficulty in the autonomous exploration process of mobile robots in a large-scale unknown environment, the present invention utilizes a path analysis module, namely the selection of extended nodes, to provide a multi-dimensional view comprehensive evaluation mechanism, effectively improving the efficiency of the mobile robot in planning the exploration path during the exploration process. At the same time, it effectively reduces the repeated exploration process of the same area, greatly improving the autonomous exploration efficiency of the mobile robot. The path tracking module proposes a pure path tracking method based on the line-of-sight distance, using the straight-line distance from the current position of the robot to the target point to adjust the speed and minimum turning radius of the robot, effectively improving the smoothness and accuracy of the mobile robot path tracking and reducing the lateral error of the tracking. Finally, through the cooperation of each module, the technical problems such as low efficiency and high consumption existing in the prior art in the autonomous exploration of ground mobile robots in an unknown environment are solved.

[0109] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for autonomous exploration of a mobile robot in an unknown environment, characterized in that: The method includes: Step S1: obtaining the real-time positioning information and point cloud information of the robot, registering the point cloud information, and updating the 3D grid map represented by the three-dimensional space using the point cloud data information; Step S2: performing incremental boundary detection in the updated 3D grid map, sampling candidate viewpoints, and evaluating the sampled viewpoints to select nodes to be expanded; Step S3: In the area detected by the robot where it can move freely, the search-based HybridA* path planning algorithm is used to search for paths from the current robot position to all candidate viewpoints obtained based on sampling, and the paths are maintained in the topology map; Step S4: Use a pure path tracking method based on dynamic ray distance to track the path from the robot's current position to the extended node.

2. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 1, characterized in that: The step S1 specifically includes: Step S11: Use a laser radar-based SLAM module to scan the environment, obtain the robot's positioning and point cloud information, and process the point cloud information to convert it into a grid map representation to update the 3D grid map in real time; Step S12: Each node in the 3D grid map corresponds to a cubic area in the environment space. When new point cloud data arrives, the 3D grid map will update the state of the corresponding cubic area. In the initial state, all areas are marked as unknown space S uk ; As the robot moves and the SLAM module scans, the unknown area will be gradually explored and updated to a known state. The known state is divided into a known free space S free and the known occupied space S occ , the free space S is known free It is the area where the robot can move freely detected by the sensor; the occupied space S is known. occ It is an inaccessible area detected by the robot through sensors.

3. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 2, characterized in that: The step S2 specifically includes: Step S21, performing boundary detection according to the updated 3D grid map, and only detecting the newly added grids recorded in the 3D grid map and the existing boundaries of the robot; Step S22, uniform random sampling is performed around the frontier boundary detected in the area where the robot can move freely through the sensor to obtain candidate viewpoints; Step S23: Perform multi-dimensional viewpoint comprehensive evaluation on the randomly sampled candidate viewpoints to screen out suitable candidate viewpoints.

4. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 3, characterized in that: The boundary in step S21 is: in the known 3D space, if there is a free grid with at least one unknown neighbor grid of the same height, then the free grid is defined as the boundary; let p i is the i-th position of the robot, F pi For robots in p i The global boundary at For p i to p i+1 The boundaries within the robot's range, For robots from p i Move to p i+1 The updated sensor senses a free grid within the range, then Boundary detection is performed in .

5. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 3, characterized in that: The extended nodes selected in step S23 are suitable candidate viewpoints selected by multi-dimensional viewpoint comprehensive weight calculation, and their weights include information gain, travel cost and topological features, as follows: The information gain is evaluated by comprehensively considering the number of unknown grids visible from the viewpoint. The calculation is performed by simulating the ray casting method of the lidar. A simulated laser ray is emitted from the viewpoint to the detected frontier boundary. The formula Calculate the number of grids covering the unknown area, where δ i represents the contribution value of the i-th unknown grid, and P(v) represents the information gain of viewpoint v; The travel cost is based on the topological structure of the map, including the actual shortest path length L(v0,v) from the current node to the candidate viewpoint and the number of nodes L passed by the path. t (v0,v), where the shortest path length is searched using the HybridA* algorithm; The topological feature explores the endpoints of the topological map based on the position of the viewpoint in the topological map, where the endpoints of the topological map represent a complete area to be explored, and the nodes are the intersections of different areas; Therefore, the viewpoint weight calculation formula is: Among them, U(v k ) is the viewpoint v k The utility of G(v k ) is the penalty function, P(v k ) is the information gain, c is the penalty factor, L(v0→v k ) is the distance from the current viewpoint to the viewpoint v k moving distance.

6. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 5, characterized in that: The pure path tracking method based on dynamic ray distance in S4 tracks the predetermined path by controlling the linear speed and steering angle of the robot, and the specific steps are as follows: Step S41, calculating the ray distance and the required turning angle according to the current position of the robot, the target position and the next target point on the path; Step S42, calculate the ray distance. The ray distance refers to the straight-line distance from the current position of the robot to the target point. The ray distance is dynamically adjusted according to the speed and minimum turning radius of the robot. The calculation formula is: d =k*V+C, where k is the speed coefficient, V is the speed of the robot, and C is the minimum turning radius of the robot; Step S43, calculating the steering angle, the steering angle θ is calculated by the Ackermann steering model, and depends on the wheelbase L and the turning radius R of the vehicle, and the calculation formula is: tanθ=L / R; Step S44: Use the PID algorithm to adjust the speed of the robot to ensure that the robot travels along the predetermined path.

7. The method for autonomous exploration of a mobile robot in an unknown environment according to claim 5, characterized in that: During the path tracking process of step S44, the position error of the robot is used to adjust the steering angle to reduce the movement deviation of the robot and keep it on the path; wherein the position error is the lateral error e(t).

8. A mobile robot autonomous exploration system in an unknown environment, characterized in that: include: The SLAM module is used to receive the robot's positioning information and the real-time point cloud information around the robot body collected by the robot's body sensor, align the point cloud information, perform point cloud data modeling, and build and update the 3D grid map; The path analysis module is used to receive the 3D grid map information transmitted by the SLAM module, perform map frontier detection, generate candidate viewpoints based on sampling at the map frontier, and screen out expansion nodes through a multi-dimensional viewpoint comprehensive evaluation mechanism; The path planning module generates the path from the robot's current position to all candidate viewpoints through the search-based HybridA* path planning algorithm, and maintains the path in the topology map; Path tracking module, the robot screens out extended nodes based on the path analysis module, obtains the path through the extended nodes through the path planning module, and tracks the path using a pure path tracking method based on ray distance.

Images