A mobile robot autonomous navigation system

Abstract

The present invention discloses an autonomous navigation system for a mobile robot, belonging to the field of self-service navigation. The system includes: an acquisition module, a multi-layer planner initialization module, a navigation instruction analysis module, and a drive module; the acquisition module is used to import sensor data and prior knowledge; the multi-layer planner initialization module is used to initialize the multi-layer planner and build a correlation matrix for the multi-layer planner; the navigation instruction analysis module is used to receive navigation instructions and perform route planning for the acquired navigation instructions; the drive module is used to perform operational control on the received route planning and adjust the state of the mobile robot at the same time. The present invention provides an autonomous navigation system for a mobile robot, which uses a perception module to perceive the environment and prior knowledge, utilizes a multi-layer multi-mode planning module to provide an execution strategy for the autonomous mobile robot, and introduces a state machine to ensure the stability of the system and the correctness of the decision.

Description

Technical Field

[0001] The present invention relates to the technical field of robot applications, and in particular to an autonomous navigation system for a mobile robot. Background Art

[0002] With the development of autonomous mobile robots in fields such as industrial automation, logistics, intelligent manufacturing, services, healthcare, firefighting, and cleaning, they are gradually gaining widespread acceptance and adoption due to their high degree of intelligence and autonomy. Autonomous mobile robots possess a high degree of autonomy, capable of autonomously moving and performing tasks within a certain range, with minimal human intervention. Autonomous mobile robot autonomous navigation systems enable autonomous mobile robots to autonomously move and perform tasks within known, unstructured environments.

[0003] In some early robotic navigation systems, the robot's navigation system first constructed a trajectory guide (such as a magnetic strip, ribbon, or QR code) around the entire workspace. The trajectory guide connected each work point, and the robot then walked along the trajectory guide. The entire process was time-consuming and labor-intensive to set up. Furthermore, the manually laid trajectory guide could not meet the optimal strategy. Moreover, the robot could only follow a preset trajectory and strategy, lacking flexibility and autonomy.

[0004] With the development of autonomous mobile robots, path planners and cost maps have been introduced to fully consider the robot's autonomy. The cost map is driven by sensors, and the path planner uses the cost map as input. Robots can perceive their surroundings through sensors (such as cameras, radar, and sonar) and make real-time decisions through the path planner. Compared to early manual decision-making, this approach greatly improves the robot's autonomy and flexibility. In complex unstructured scenarios, due to factors such as the limitations of sensor perception and the incompleteness of the planner, the robot's behavioral decisions often have local optimality and incompleteness, which limits its use.

[0005] Therefore, people need a mobile robot autonomous navigation system to solve the above problems. Summary of the Invention

[0006] The object of the present invention is to provide an autonomous navigation system for a mobile robot to solve the problems raised in the above background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] A mobile robot autonomous navigation system, the system comprising: an acquisition module, a multi-layer planner initialization module, a navigation instruction analysis module and a driving module;

[0009] The acquisition module is used to import sensor data and prior knowledge;

[0010] The multi-layer planner initialization module is used to initialize the multi-layer planner and build the relevance matrix of the multi-layer planner;

[0011] The navigation instruction analysis module is used to receive navigation instructions and perform route planning for the obtained navigation instructions;

[0012] The driving module is used to control the operation of the received route plan and adjust the state of the mobile robot;

[0013] The output end of the acquisition module is connected to the input end of the multi-layer planner initialization module; the output end of the multi-layer planner initialization module is connected to the input end of the navigation instruction analysis module; the output end of the acquisition module is connected to the input end of the navigation instruction analysis module; the output end of the navigation instruction analysis module is connected to the input end of the driving module.

[0014] According to the above technical solution, the acquisition module includes a sensor data import unit and a priori knowledge import unit;

[0015] The sensor data import unit is used to import data from photoelectric switches, anti-collision strips, cameras, laser radars, and sonars;

[0016] The prior knowledge importing unit is used to import prior maps, map annotations and robot information.

[0017] According to the above technical solution, the multi-layer planner initialization module includes an upper-layer planner unit, a middle-layer planner unit, and a bottom-layer planner unit;

[0018] The upper planner unit is used to initialize the upper planner and initialize the upper planner relevance matrix;

[0019] The intermediate-layer planner unit is used to initialize the intermediate-layer planner and initialize the intermediate-layer planner relevance matrix;

[0020] The bottom-level planner unit is used to initialize the bottom-level planner and initialize the bottom-level planner association matrix.

[0021] According to the above technical solution, the navigation instruction analysis module includes a navigation instruction acquisition unit and a navigation instruction planning unit;

[0022] The navigation instruction acquisition unit is used to convert the sensor data into the reference coordinate system of the robot in accordance with the coordinate transformation relationship in the robot description file to obtain the navigation instruction;

[0023] The navigation instruction planning unit is used to plan a route for the acquired navigation instruction.

[0024] According to the above technical solution, the driving module includes an operation control unit and a state adjustment unit;

[0025] The operation control unit is used to issue control instructions to the autonomous mobile robot to implement the route planned by the navigation instruction planning unit;

[0026] The state adjustment unit is used to make adjustments according to the system working state of the autonomous mobile robot.

[0027] According to the above technical solution, the sensor data import unit includes the data import of photoelectric switch, anti-collision strip, camera, laser radar and sonar, wherein: the photoelectric switch creates a point cloud set of obstacles at the corresponding position according to the position of the photoelectric switch, wherein the reference coordinate system of the point cloud is the reference coordinate of the photoelectric switch, and the point cloud coordinate is p opt =(l opt ,0,0),l opt is the effective detection distance of the photoelectric switch, which is manually input according to the sensor parameters; the anti-collision bar is manually input according to the installation position of the anti-collision bar, and a point cloud set of obstacles at corresponding positions and angles is created, where the reference coordinate system of the point cloud is the base coordinate of the anti-collision bar, and the point cloud coordinates are (0,0,0); the camera needs to sample the depth camera, segment the invalid point cloud by extracting different objects in the point cloud, thereby achieving divide and conquer, highlighting the key points and processing them separately (the specific segmentation algorithm belongs to the existing technology and is not involved in this invention) to obtain the processed point cloud set; the laser radar is manually input based on the laser radar scanning data, combined with the laser radar detection angle range and angular resolution, and converts the scanning point set into a point cloud set in the laser radar coordinate system. Assume that the scanning range of the laser radar is [α,β] and the angular resolution of the laser radar is dθ, then the coordinates of the scanning point (x i ,y i ) can be expressed as:

[0028] x i =d i cos(α+idθ)

[0029] y i =d i sin(α+idθ)

[0030] Where α+idθ<β; the sonar generates an arc-shaped point cloud set at the corresponding position according to the sonar detection range and the distance to the detected obstacle. The arc radius is equal to the obstacle distance, and the maximum number of point clouds is generally set to 10.

[0031] According to the above technical solution, the prior knowledge import unit imports priority data of virtual wall areas, forbidden areas, priority areas, feasible areas, unknown areas and functional areas. The virtual wall area is treated as a wall, and the virtual wall area and its edge expansion area belong to the forbidden area, and the autonomous mobile robot cannot pass through; the autonomous mobile robot cannot pass through the forbidden area; the autonomous mobile robot has priority passing right in the priority area, that is, the autonomous mobile robot gives priority to passing through the priority area, and its passing priority is the highest; the feasible area is a general pass area, and its priority is second only to the priority area; the autonomous mobile robot can also pass through the unknown area, and its priority is second only to the feasible area; in the functional area, the autonomous mobile robot will trigger preset related functions.

[0032] According to the above technical solution, the upper-level planner unit is divided into initializing the upper-level planner and initializing the upper-level planner association matrix. Initializing the upper-level planner is to switch the system state to "preparing" and import the upper-level planner set: in, The topology planner takes as input a manually annotated path topology map. The stopping points of the autonomous mobile robot are nodes of the topology map. The weights between the nodes are the lengths of the annotated paths. The optimal path is derived through the shortest path algorithm. For an autonomous planner based on a search strategy, the specific implementation process of the shortest path algorithm is as follows:

[0033] (1) Create an adjacency matrix A∈R based on nodes and weights n×n , n is the number of nodes, a ij is the path length from node i to node j. If the two nodes are not connected, the weight is set to infinity;

[0034] (2) Take the starting node from the original set V and add it to the vertex set T;

[0035] (3) Traverse the vertex set T, take out the path length from the selected node to the remaining vertices in V from the adjacency matrix A, and sum it with the shortest path length of the selected node to obtain the alternative path and its corresponding path length;

[0036] (4) Select the shortest path from the alternative paths, remove the end point of the path from the original set V and add it to the vertex set T, and save the length of the shortest path to the node;

[0037] If the node newly added to the vertex set is the target point, exit; otherwise, return to (3);

[0038] Initialize the upper planner correlation matrix, denoted as C 0 :

[0039]

[0040] C 0 is the upper planner set and the upper prior knowledge set The correlation degree of Represents the upper-level planner and prior knowledge The intermediate planner unit needs to import the intermediate planner set for initializing the intermediate planner and autonomous mobile robot motion primitives; initialize the middle layer planner correlation matrix C 1 :

[0041]

[0042] C 1 is the set of intermediate layer planners and the set of intermediate layer prior knowledge The correlation degree of Representation Planner and prior knowledge The bottom-level planner unit, regarding initialization of the bottom-level planner, needs to import the bottom-level planner set Initialize the underlying planner correlation matrix C 2 :

[0043]

[0044] C 2 is the underlying planner set and the underlying prior knowledge set The correlation degree of Representation Planner and prior knowledge degree of correlation.

[0045] According to the above technical solution, the navigation instruction planning unit receives the obtained navigation instruction. If the navigation instruction is invalid, the system state is switched to "idle". If the navigation instruction is valid, the upper planner thread is opened, and the upper reward function combines the navigation instruction information and prior knowledge to give the upper planner set P 0 The planner scores in , where the prior knowledge X 0 Includes: Navigation system status Autonomous mobile robot system status Last planner status feedback Planner Priorities and manually annotated paths Generally, When there is a manually marked path When no manual path is marked Reward R 0The highest planner will be selected as the target planner to perform the upper-level path planning, namely: When initializing C 0 hour, and The corresponding correlation should be set to the maximum reward value. The upper planner plans the upper path in combination with the global cost map and switches the system state to "planning". If the upper planner plans the time t t Or planning times N t If the value exceeds the preset value, an error is thrown, the global cost map is reset, and the system state is adjusted to "upper planner planning failure"; (a) Upper path inspection and preprocessing, starting from the starting point, traversing the points on the upper path, and calculating the distance between the path point and the starting point Where (x, y) is the coordinate of the current path point, (x0, y0) is the coordinate of the starting point, and combined with the global preview distance L, if l>L, this point is selected as the global preview point. Usually, L is equal to half the width of the local cost map;

[0046] (b) Open the middle-layer planner thread, and the middle-layer reward function combines the navigation instructions and prior knowledge to give the middle-layer planner set P 1 The intermediate planner scores in which the prior knowledge X 1 Including autonomous mobile robot system status Last planner status feedback Planner Priorities and navigation system status Reward R 1 The highest planner will be selected as the target planner to perform the intermediate layer path planning, that is: (c) The middle-level planner combines the local cost map and the autonomous mobile robot motion primitives to plan the middle-level path;

[0047] (d) If the middle-level planner plans time t m Or planning times N m Exceeds the preset value, where t m and N m If the navigation command is passed in, an error is thrown and the local cost map is reset, and the system state is adjusted to "intermediate planner planning failure";

[0048] (e) Open the bottom-level planner thread, and the bottom-level reward function combines the navigation instructions and prior knowledge to give the bottom-level planner set P 2 The planner scores in , where the prior knowledge X 2 Including autonomous mobile robot system status Last planner status feedback Planner Priorities Distance level from target point and navigation system status in, The autonomous mobile robot is divided into different levels according to the location of the global preview point. Generally, we divide it into three levels:

[0049] (i) Global preview point (x L ,y L ) does not coincide with the end point and the distance from the preview point to the starting point When the global preview distance is less than L

[0050] (ii) When the distance from the preview point to the starting point is greater than the global preview distance L and the global preview point and the end point do not coincide

[0051] (iii) When the global preview point and the end point coincide Reward R 2 The highest planner will be selected as the target planner for this path planning:

[0052] (f) The bottom-level planner plans the local path based on the local cost map and issues control instructions for the autonomous mobile robot.

[0053] According to the above technical solution, the state adjustment unit determines the current position of the autonomous mobile robot. If the autonomous mobile robot receives a valid control instruction, the system state is switched to "navigating"; if it reaches the target point, the system state is adjusted to "navigation completed". After this state lasts for a certain period of time, the system state returns to "idle".

[0054] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0055] 1. This invention takes the autonomy of the robot into consideration and introduces a path planner and a cost map. The cost map is driven by sensors, and the path planner uses the cost map as input. The robot perceives the surrounding environment through sensors (such as cameras, radar, sonar, etc.) and makes real-time decisions through the path planner. Compared with early manual decision-making, the robot's autonomy is greatly improved and it is more flexible.

[0056] 2. The present invention proposes an automatic navigation system for an autonomous mobile robot. The environmental perception module incorporates a classic two-layer costmap framework, namely a global costmap and a local costmap. Unlike traditional sensor-driven hierarchical costmaps, the environmental perception module of the present invention utilizes a hierarchical costmap driven by both prior knowledge and sensor data, taking into account the incompleteness of sensor detection. This effectively overcomes misperceptions and underperceptions caused by sensor measurement errors and perception incompleteness. The planning module of the present invention adopts a multi-layer, multi-mode design, hierarchically divided into an upper-layer planner, a middle-layer planner, and a bottom-layer planner. To ensure the continuity of decision-making actions, each layer of the multi-layer planner adopts an independent thread operation mode, effectively overcoming the lag problem caused by the response time of the planners at each layer. Considering the incompleteness of a single planner, each layer of the planning module of the present invention can simultaneously integrate multiple planners and switch modes through a reward function composed of state, perception, and prior knowledge, effectively overcoming the incompleteness of a single planner. In addition, the present invention rationally utilizes a state machine to transition system states, further ensuring the stability of the overall system and the accuracy of decision-making.

[0057] 3. The automatic navigation system of the present invention can effectively integrate various sensors and prior knowledge to ensure the completeness of perception. At the same time, it adopts a multi-layer multi-mode planner design, can integrate various types of planners, and make decisions through reward functions. Its deployment and use are more flexible, and it is more adaptable to scenarios and can cover more usage scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0059] Figure 1 It is a structural schematic diagram of a mobile robot autonomous navigation system of the present invention;

[0060] Figure 2 It is a flow chart of a mobile robot autonomous navigation system of the present invention. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 making creative efforts are within the scope of protection of the present invention.

[0062] See Figure 1-Figure 2, in this embodiment: a mobile robot autonomous navigation system, the system includes: an acquisition module, a multi-layer planner initialization module, a navigation instruction analysis module and a driving module;

[0063] The acquisition module is used to import sensor data and prior knowledge;

[0064] The multi-layer planner initialization module is used to initialize the multi-layer planner and build the relevance matrix of the multi-layer planner;

[0065] The navigation instruction analysis module is used to receive navigation instructions and perform route planning for the obtained navigation instructions;

[0066] The driving module is used to control the operation of the received route plan and adjust the state of the mobile robot;

[0067] The output end of the acquisition module is connected to the input end of the multi-layer planner initialization module; the output end of the multi-layer planner initialization module is connected to the input end of the navigation instruction analysis module; the output end of the acquisition module is connected to the input end of the navigation instruction analysis module; the output end of the navigation instruction analysis module is connected to the input end of the driving module.

[0068] According to the above technical solution, the acquisition module includes a sensor data import unit and a priori knowledge import unit;

[0069] The sensor data import unit is used to import data from photoelectric switches, anti-collision strips, cameras, laser radars, and sonars;

[0070] The prior knowledge importing unit is used to import prior maps, map annotations and robot information.

[0071] According to the above technical solution, the multi-layer planner initialization module includes an upper-layer planner unit, a middle-layer planner unit, and a bottom-layer planner unit;

[0072] The upper planner unit is used to initialize the upper planner and initialize the upper planner relevance matrix;

[0073] The intermediate-layer planner unit is used to initialize the intermediate-layer planner and initialize the intermediate-layer planner relevance matrix;

[0074] The bottom-level planner unit is used to initialize the bottom-level planner and initialize the bottom-level planner association matrix.

[0075] According to the above technical solution, the navigation instruction analysis module includes a navigation instruction acquisition unit and a navigation instruction planning unit;

[0076] The navigation instruction acquisition unit is used to convert the sensor data into the reference coordinate system of the robot in accordance with the coordinate transformation relationship in the robot description file to obtain the navigation instruction;

[0077] The navigation instruction planning unit is used to plan a route for the acquired navigation instruction.

[0078] According to the above technical solution, the driving module includes an operation control unit and a state adjustment unit;

[0079] The operation control unit is used to issue control instructions to the autonomous mobile robot to implement the route planned by the navigation instruction planning unit;

[0080] The state adjustment unit is used to make adjustments according to the system working state of the autonomous mobile robot.

[0081] According to the above technical solution, the sensor data import unit includes the data import of photoelectric switch, anti-collision strip, camera, laser radar and sonar, wherein: the photoelectric switch creates a point cloud set of obstacles at the corresponding position according to the position of the photoelectric switch, wherein the reference coordinate system of the point cloud is the reference coordinate of the photoelectric switch, and the point cloud coordinate is p opt =(l opt ,0,0),l opt is the effective detection distance of the photoelectric switch, which is manually input according to the sensor parameters; the anti-collision bar is manually input according to the installation position of the anti-collision bar, and a point cloud set of obstacles at the corresponding position angle is created, where the reference coordinate system of the point cloud is the base coordinate of the anti-collision bar, and the point cloud coordinates are (0,0,0); the camera needs to sample the depth camera and segment the invalid point cloud (the specific segmentation algorithm belongs to the existing technology and is not involved in this invention) to obtain the processed point cloud set; the laser radar is manually input based on the laser radar scanning data, combined with the laser radar detection angle range and angular resolution, and the scanning point set is converted into a point cloud set in the laser radar coordinate system. Assume that the scanning range of the laser radar is [α,β] and the angular resolution of the laser radar is dθ, then the coordinates of the scanning point (x i ,y i ) can be expressed as:

[0082] x i =d i cos(α+idθ)

[0083] y i =d i sin(α+idθ)

[0084] Where α+idθ<β; the sonar generates an arc-shaped point cloud set at the corresponding position according to the sonar detection range and the distance to the detected obstacle. The arc radius is equal to the obstacle distance, and the maximum number of point clouds is generally set to 10.

[0085] According to the above technical solution, the prior knowledge import unit imports priority data of virtual wall areas, forbidden areas, priority areas, feasible areas, unknown areas and functional areas. The virtual wall area is treated as a wall, and the virtual wall area and its edge expansion area belong to the forbidden area, and the autonomous mobile robot cannot pass through; the autonomous mobile robot cannot pass through the forbidden area; the autonomous mobile robot has priority passing right in the priority area, that is, the autonomous mobile robot gives priority to passing through the priority area, and its passing priority is the highest; the feasible area is a general pass area, and its priority is second only to the priority area; the autonomous mobile robot can also pass through the unknown area, and its priority is second only to the feasible area; in the functional area, the autonomous mobile robot will trigger preset related functions.

[0086] According to the above technical solution, the upper-level planner unit is divided into initializing the upper-level planner and initializing the upper-level planner association matrix. Initializing the upper-level planner is to switch the system state to "preparing" and import the upper-level planner set: in, The topology planner takes as input a manually annotated path topology map. The stopping points of the autonomous mobile robot are nodes of the topology map. The weights between the nodes are the lengths of the annotated paths. The optimal path is derived through the shortest path algorithm. For an autonomous planner based on a search strategy, the specific implementation process of the shortest path algorithm is as follows:

[0087] (1) Create an adjacency matrix A∈R based on nodes and weights n×n , n is the number of nodes, a ij is the path length from node i to node j. If the two nodes are not connected, the weight is set to infinity;

[0088] (2) Take the starting node from the original set V and add it to the vertex set T;

[0089] (3) Traverse the vertex set T, take out the path length from the selected node to the remaining vertices in V from the adjacency matrix A, and sum it with the shortest path length of the selected node to obtain the alternative path and its corresponding path length;

[0090] (4) Select the shortest path from the alternative paths, remove the end point of the path from the original set V and add it to the vertex set T, and save the length of the shortest path to the node;

[0091] If the node newly added to the vertex set is the target point, exit; otherwise, return to (3);

[0092] Initialize the upper planner correlation matrix, denoted as C 0 :

[0093]

[0094] C 0 is the upper planner set and the upper prior knowledge set The correlation degree of Represents the upper-level planner and prior knowledge The intermediate planner unit needs to import the intermediate planner set for initializing the intermediate planner and autonomous mobile robot motion primitives; initialize the middle layer planner correlation matrix C 1 :

[0095]

[0096] C 1 is the set of intermediate layer planners and the set of intermediate layer prior knowledge The correlation degree of Representation Planner and prior knowledge The bottom-level planner unit, regarding initialization of the bottom-level planner, needs to import the bottom-level planner set Initialize the underlying planner correlation matrix C 2 :

[0097]

[0098] C 2 is the underlying planner set and the underlying prior knowledge set The correlation degree of Representation Planner and prior knowledge degree of correlation.

[0099] According to the above technical solution, the navigation instruction planning unit receives the obtained navigation instruction. If the navigation instruction is invalid, the system state is switched to "idle". If the navigation instruction is valid, the upper planner thread is opened, and the upper reward function combines the navigation instruction information and prior knowledge to give the upper planner set P 0 The planner scores in , where the prior knowledge X 0 Includes: Navigation system status Autonomous mobile robot system status Last planner status feedback Planner Priorities and manually annotated paths Generally, When there is a manually marked path When no manual path is marked Reward R0 The highest planner will be selected as the target planner to perform the upper-level path planning, namely: When initializing C 0 hour, and The corresponding correlation should be set to the maximum reward value. The upper planner plans the upper path in combination with the global cost map and switches the system state to "planning". If the upper planner plans the time t t Or planning times N t If the value exceeds the preset value, an error is thrown, the global cost map is reset, and the system state is adjusted to "upper planner planning failure"; (a) Upper path inspection and preprocessing, starting from the starting point, traversing the points on the upper path, and calculating the distance between the path point and the starting point Where (x, y) is the coordinate of the current path point, (x0, y0) is the coordinate of the starting point, and combined with the global preview distance L, if l>L, this point is selected as the global preview point. Usually, L is equal to half the width of the local cost map;

[0100] (b) Open the middle-layer planner thread, and the middle-layer reward function combines the navigation instructions and prior knowledge to give the middle-layer planner set P 1 The intermediate planner scores in which the prior knowledge X 1 Including autonomous mobile robot system status Last planner status feedback Planner Priorities and navigation system status Reward R 1 The highest planner will be selected as the target planner to perform the intermediate layer path planning, that is: (c) The middle-level planner combines the local cost map and the autonomous mobile robot motion primitives to plan the middle-level path;

[0101] (d) If the middle-level planner plans time t m Or planning times N m Exceeds the preset value, where t m and N m If the navigation command is passed in, an error is thrown and the local cost map is reset, and the system state is adjusted to "intermediate planner planning failure";

[0102] (e) Open the bottom-level planner thread, and the bottom-level reward function combines the navigation instructions and prior knowledge to give the bottom-level planner set P 2 The planner scores in , where the prior knowledge X 2 Including autonomous mobile robot system status Last planner status feedback Planner Priorities Distance level from target point and navigation system status in, The autonomous mobile robot is divided into different levels according to the location of the global preview point. Generally, we divide it into three levels:

[0103] (i) Global preview point (x L ,y L ) does not coincide with the end point and the distance from the preview point to the starting point When the global preview distance is less than L

[0104] (ii) When the distance from the preview point to the starting point is greater than the global preview distance L and the global preview point and the end point do not coincide

[0105] (iii) When the global preview point and the end point coincide Reward R 2 The highest planner will be selected as the target planner for this path planning:

[0106] (f) The bottom-level planner plans the local path based on the local cost map and issues control instructions for the autonomous mobile robot.

[0107] According to the above technical solution, the state adjustment unit determines the current position of the autonomous mobile robot. If the autonomous mobile robot receives a valid control instruction, the system state is switched to "navigating"; if it reaches the target point, the system state is adjusted to "navigation completed". After this state lasts for a certain period of time, the system state returns to "idle".

[0108] Embodiment 1: Initialize the upper-layer path planner and import the upper-layer path planner set, which specifically includes an autonomous planner based on a search strategy and a topology planner based on prior knowledge; initialize the middle-layer path planner, import the middle-layer path planner set and load the autonomous mobile robot motion primitives, the middle-layer planner is composed of an autonomous planner based on motion primitives; initialize the bottom-layer planner and import the bottom-layer planner set, the bottom-layer planner provides a control strategy for the autonomous mobile robot; the autonomous mobile robot receives a navigation instruction, opens the upper-layer planner thread, and plans the upper-layer path in combination with the global cost map; checks the upper-layer path, and calculates the global preview point based on the global preview distance; opens the middle-layer path planner thread, plans the middle-layer path in combination with the local cost map and the autonomous mobile robot motion primitives; opens the bottom-layer planner thread, plans the autonomous mobile robot control trajectory in combination with the local cost map and the autonomous mobile robot state, and issues and implements the autonomous mobile robot control instruction.

[0109] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0110] 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 aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A mobile robot autonomous navigation system, characterized by: The system includes: an acquisition module, a multi-layer planner initialization module, a navigation instruction analysis module and a driving module; The acquisition module is used to import sensor data and prior knowledge; The multi-layer planner initialization module is used to initialize the multi-layer planner and build the relevance matrix of the multi-layer planner; The upper-level planner unit is divided into initializing the upper-level planner and initializing the upper-level planner association matrix. Initializing the upper-level planner switches the system state to "preparing" and imports the upper-level planner set: in, The topology planner takes as input a manually annotated path topology map. The stopping points of the autonomous mobile robot are nodes of the topology map. The weights between the nodes are the lengths of the annotated paths. The optimal path is derived through the shortest path algorithm. For an autonomous planner based on a search strategy, the specific implementation process of the shortest path algorithm is as follows: (1) Create an adjacency matrix A∈R based on nodes and weights n×n , n is the number of nodes, a ij is the path length from node i to node j. If the two nodes are not connected, the weight is set to infinity; (2) Take the starting node from the original set V and add it to the vertex set T; (3) Traverse the vertex set T, take out the path length from the selected node to the remaining vertices in V from the adjacency matrix A, and sum it with the shortest path length of the selected node to obtain the alternative path and its corresponding path length; (4) Select the shortest path from the alternative paths, remove the end point of the path from the original set V and add it to the vertex set T, and save the length of the shortest path to the node; If the node newly added to the vertex set is the target point, exit; otherwise, return to (3); Initialize the upper planner correlation matrix, denoted as C 0 : C 0 is the upper planner set and the upper prior knowledge set The correlation degree of Represents the upper-level planner and prior knowledge The degree of association; the middle-level planner unit, about initializing the middle-level planner, you need to import the middle-level planner set and autonomous mobile robot motion primitives; initialize the middle layer planner correlation matrix C 1 : C 1 is the set of intermediate layer planners and the set of intermediate layer prior knowledge The correlation degree of Representation Planner and prior knowledge The correlation degree of the bottom planner unit, about initializing the bottom planner, you need to import the bottom planner set Initialize the underlying planner correlation matrix C 2 : C 2 is the underlying planner set and the underlying prior knowledge set The correlation degree of Representation Planner and prior knowledge degree of relevance; The navigation instruction analysis module is used to receive navigation instructions and perform route planning for the obtained navigation instructions; The driving module is used to control the operation of the received route plan and adjust the state of the mobile robot; The output end of the acquisition module is connected to the input end of the multi-layer planner initialization module; the output end of the multi-layer planner initialization module is connected to the input end of the navigation instruction analysis module; the output end of the acquisition module is connected to the input end of the navigation instruction analysis module; the output end of the navigation instruction analysis module is connected to the input end of the driving module.

2. The mobile robot autonomous navigation system according to claim 1, characterized in that: The acquisition module includes a sensor data import unit and a priori knowledge import unit; The sensor data import unit is used to import data from photoelectric switches, anti-collision strips, cameras, laser radars, and sonars; The prior knowledge importing unit is used to import prior maps, map annotations and robot information.

3. The mobile robot autonomous navigation system according to claim 1, characterized in that: The multi-layer planner initialization module includes an upper-layer planner unit, a middle-layer planner unit, and a bottom-layer planner unit; The upper planner unit is used to initialize the upper planner and initialize the upper planner relevance matrix; The intermediate-layer planner unit is used to initialize the intermediate-layer planner and initialize the intermediate-layer planner relevance matrix; The bottom-level planner unit is used to initialize the bottom-level planner and initialize the bottom-level planner association matrix.

4. The mobile robot autonomous navigation system according to claim 1, characterized in that: The navigation instruction analysis module includes a navigation instruction acquisition unit and a navigation instruction planning unit; The navigation instruction acquisition unit is used to convert the sensor data into the reference coordinate system of the robot in accordance with the coordinate transformation relationship in the robot description file to obtain the navigation instruction; The navigation instruction planning unit is used to plan a route for the acquired navigation instruction.

5. The mobile robot autonomous navigation system according to claim 1, characterized in that: The driving module includes an operation control unit and a state adjustment unit; The operation control unit is used to issue control instructions to the autonomous mobile robot to implement the route planned by the navigation instruction planning unit; The state adjustment unit is used to make adjustments according to the system working state of the autonomous mobile robot.

6. The mobile robot autonomous navigation system according to claim 2, characterized in that: The sensor data import unit includes data import of photoelectric switches, anti-collision strips, cameras, laser radars and sonars, wherein: the photoelectric switch creates a point cloud set of obstacles at the corresponding position according to the position of the photoelectric switch, wherein the reference coordinate system of the point cloud is the reference coordinate of the photoelectric switch, and the point cloud coordinate is p opt =(l opt ,0,0),l opt is the effective detection distance of the photoelectric switch, which is manually input according to the sensor parameters; the anti-collision bar is manually input according to the installation position of the anti-collision bar, and a point cloud set of obstacles at the corresponding position angle is created, where the reference coordinate system of the point cloud is the base coordinate of the anti-collision bar, and the point cloud coordinate is (0,0,0); the camera needs to sample the depth camera and segment the invalid point cloud to obtain the processed point cloud set; the laser radar is manually input according to the laser radar scanning data, combined with the laser radar detection angle range and angular resolution, and the scanning point set is converted into a point cloud set in the laser radar coordinate system. Assume that the scanning range of the laser radar is [α,β] and the angular resolution of the laser radar is dθ, then the coordinates of the scanning point (x i ,y i ) is expressed as: x i =d i cos(α+idθ) y i =d i sin(α+idθ) Where α+idθ<β; the sonar generates an arc-shaped point cloud set at the corresponding position based on the sonar detection range and the distance to the detected obstacle. The arc radius is equal to the obstacle distance, and the maximum number of point clouds is set to 10.

7. The mobile robot autonomous navigation system according to claim 2, characterized in that: The prior knowledge import unit imports priority data of virtual wall areas, forbidden areas, priority areas, feasible areas, unknown areas and functional areas, the virtual wall area is treated as a wall, the virtual wall area and its edge expansion area are both forbidden areas, and the autonomous mobile robot cannot pass through; the autonomous mobile robot cannot pass through the forbidden area; the autonomous mobile robot has priority passing right in the priority area, that is, the autonomous mobile robot gives priority to passing through the priority area, and its passing priority is the highest; the feasible area is a generally passable area, and its priority is second only to the priority area; the autonomous mobile robot can also pass through the unknown area, and its priority is second only to the feasible area; in the functional area, the autonomous mobile robot will trigger preset related functions.

8. The mobile robot autonomous navigation system according to claim 4, characterized in that: The navigation instruction planning unit receives the acquired navigation instruction and switches the system state to "idle" if the navigation instruction is invalid; If the navigation instruction is valid, open the upper planner thread, and the upper reward function combines the navigation instruction information and prior knowledge to give the upper planner set P 0 The planner scores in , where the prior knowledge X 0 Includes: Navigation system status Autonomous mobile robot system status Last planner status feedback Planner Priorities and manually annotated paths Generally, When there is a manually marked path When no manual path is marked Reward R 0 The highest planner will be selected as the target planner to perform the upper-level path planning, namely: When initializing C 0 When the topology planner and The corresponding correlation should be set to the maximum reward value. The upper planner plans the upper path in combination with the global cost map and switches the system state to "planning". If the upper planner plans the time t t Or planning times N t If it exceeds the preset value, an error is thrown, the global cost map is reset, and the system status is adjusted to "upper planner planning failure"; (a) Upper-level path inspection and preprocessing: starting from the starting point, traversing the points on the upper-level path and calculating the distance between the path points and the starting point Where (x, y) is the coordinate of the current path point, (x0, y0) is the coordinate of the starting point, and combined with the global preview distance L, if l>L, this point is selected as the global preview point. Usually, L is equal to half the width of the local cost map; (b) Open the middle-layer planner thread, and the middle-layer reward function combines the navigation instructions and prior knowledge to give the middle-layer planner set P 1 The intermediate planner scores in which the prior knowledge X 1 Including autonomous mobile robot system status Last planner status feedback Planner Priorities and navigation system status Reward R 1 The highest planner will be selected as the target planner to perform the intermediate layer path planning, that is: (c) The middle-level planner combines the local cost map and the autonomous mobile robot motion primitives to plan the middle-level path; (d) If the middle-level planner plans time t m Or planning times N m If it exceeds the preset value, an error is thrown and the local cost map is reset, and the system status is adjusted to "intermediate planner planning failure"; (e) Open the bottom-level planner thread, and the bottom-level reward function combines the navigation instructions and prior knowledge to give the bottom-level planner set P 2 The planner scores in , where the prior knowledge X 2 Including autonomous mobile robot system status Last planner status feedback Planner Priorities Distance level from target point and navigation system status in, The autonomous mobile robot is divided into different levels according to the location of the global preview point. Generally, we divide it into three levels: (i) Global preview point (x L ,y L ) does not coincide with the end point and the distance from the preview point to the starting point When the global preview distance is less than L (ii) When the distance from the preview point to the starting point is greater than the global preview distance L and the global preview point and the end point do not coincide (iii) When the global preview point and the end point coincide Reward R 2 The highest planner will be selected as the target planner for this path planning: (f) The bottom-level planner plans the local path based on the local cost map and issues control instructions for the autonomous mobile robot.

9. The mobile robot autonomous navigation system according to claim 5, characterized in that: The state adjustment unit determines the current position of the autonomous mobile robot. If the autonomous mobile robot receives a valid control instruction, the system state is switched to "navigating"; if it reaches the target point, the system state is adjusted to "navigation completed". After this state lasts for a certain period of time, the system state returns to "idle".

Images