Static path planning method of workbin robot

Through the static path planning method, the global environment model and path constraints are used to generate stable material box robot paths, solving the efficiency and stability problems of traditional dynamic path planning in complex environments, and achieving efficient and safe path planning.

CN118938923AInactive Publication Date: 2024-11-12CHONGQING SHOUHENG SOFTWARE CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411054082.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional dynamic path planning methods are susceptible to real-time obstacles and communication delays in complex warehouse environments, resulting in inefficient path planning, unstable operation of robots, and difficulty in dealing with multi-task concurrency, making it prone to path conflicts.

Method used

A static path planning method for material box robot is proposed. By reading the global environment model, analyzing the in-store instructions, searching and generating static paths, satisfying the path constraints, and ensuring the feasibility and security of the paths.

Benefits of technology

It improves the efficiency and accuracy of path planning, reduces the amount of real-time computing, avoids perception errors in dynamic path planning, solves the problem of path conflict during multi-task concurrency, and ensures the stability and security of robot operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118938923A_ABST
    Figure CN118938923A_ABST
Patent Text Reader

Abstract

The invention provides a static path planning method for warehouse-in and warehouse-out of a workbin robot, and the method comprises the steps: reading a global environment model of a warehouse where the workbin robot is located, and defining an obstacle region and a free operation region in the global environment model; analyzing a warehouse-in and warehouse-out instruction of the workbin robot to determine a current initial position and a current target point position of the workbin robot; according to the position of the starting point, the position of the current target point and a global environment model, searching for a static path of warehouse-in and warehouse-out of the workbin robot to obtain a plurality of selectable static paths so as to form a selectable static path set; and based on a set static path constraint, solving the selectable static path set so as to select a selectable static path meeting the set path constraint from the selectable static path set and use the selectable static path as a planned static path.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of the present application relate to the field of intelligent manufacturing, and in particular to a static path planning method for a material box robot. Background Art

[0002] In the field of automated warehousing, bin robots are key equipment for material handling and storage. Their efficient and accurate path planning capabilities are crucial to improving warehouse operation efficiency. Traditionally, the path planning of bin robots mainly relies on dynamic path planning methods, that is, the robot perceives the environment in real time and makes path decisions during operation. However, this method is easily affected by factors such as real-time obstacles and communication delays in complex warehouse environments, resulting in inefficient path planning, unstable robot operation, and even collision accidents.

[0003] In particular, in scenarios where frequent inbound and outbound tasks are required, dynamic path planning often fails to fully utilize prior knowledge of the warehouse environment, such as fixed obstacle locations and commonly used traffic areas, resulting in the need to re-explore the path for each task, increasing the computational burden and response time. In addition, dynamic path planning is difficult to effectively handle multi-task concurrency. When multiple bin robots are operating at the same time, path conflicts are prone to occur, affecting overall operating efficiency. Summary of the invention

[0004] The purpose of this application is to propose a static path planning method for a material box robot to solve or overcome the above-mentioned technical problems existing in the prior art.

[0005] The present application provides a static path planning method for a material box robot, which includes:

[0006] Reading a global environment model of the warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area;

[0007] Parsing the in-and-out instructions of the material box robot to determine the current starting position and the current target point position of the material box robot;

[0008] According to the starting point position, the current target point position and the global environment model, a static path for the material box robot to enter and exit the warehouse is searched to obtain a plurality of optional static paths to form an optional static path set;

[0009] Based on the set static path constraints, the optional static path set is solved to select an optional static path that meets the set path constraints and use it as the planned static path.

[0010] Optionally, the inbound and outbound instructions include inbound instructions and / or outbound instructions, so as to complete an outbound or inbound task according to the task order, and multiple goods may be taken or unloaded at one time;

[0011] The warehousing instruction is used to schedule the material box robot to place the goods from the starting position to the warehousing workbench; correspondingly, the optional static path is: the route of the material box robot from its optional starting point position to the warehousing workbench, from the warehousing workbench to the first shelf, and from the first shelf to the next shelf until all the goods are placed;

[0012] The warehousing instruction is used to schedule the material box robot to pick up goods from the starting position to the warehousing workbench; correspondingly, the optional static path includes: the route of the material box robot from the starting position to the first shelf, the route from the first shelf to the next shelf until the goods required for the order are picked up, and the route from the last shelf to the outbound workbench.

[0013] Optionally, the reading of a global environment model of a warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area, further includes:

[0014] Obtaining a global environment model file of the warehouse where the material box robot is located, parsing the global environment model file, and determining the central coordinates of obstacles in the warehouse and the outlines of the obstacles;

[0015] A global environment model of the warehouse is constructed based on the central coordinates of the obstacle and the outline of the obstacle.

[0016] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0017] Gridding the area occupied by the warehouse to obtain a plurality of grids to form a grid array;

[0018] Based on the central coordinates of the obstacle in the warehouse, matching the outline of the obstacle with the grid array to determine a mapping relationship between the obstacle and the grid array;

[0019] Based on the mapping relationship, the grid array is encoded to obtain obstacle position encoding and free area position encoding;

[0020] The obstacle position codes and free area position codes are projected onto the grid array to form a global environment model of the warehouse.

[0021] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0022] Based on the center coordinates of the obstacle and the outline of the obstacle, defining a plurality of designated position coordinates that the material box robot will pass through when traveling in the warehouse to form a designated position coordinate sequence;

[0023] Determine the route of the material box robot from its starting position point coordinates to the nearest specified position coordinates, and the route from the nearest specified position coordinates to the next specified position coordinates, iterating in sequence to obtain a plurality of routes;

[0024] Map each route as an edge and mark obstacles on each edge;

[0025] All edges loaded with obstacle identifiers are spliced ​​together in the order of iteration to form a global environment model of the warehouse.

[0026] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0027] Generating polygons representing different obstacles based on the central coordinates of the obstacles and the outlines of the obstacles;

[0028] Determine the coordinates of each vertex of each polygon to establish a line connecting the optional starting point position and all vertices, to establish a line connecting the optional target point position and all vertices, and a line connecting vertices of different polygons;

[0029] The optional starting point position, the optional target point position, the polygon, the line connecting the optional starting point position and all vertices, the line connecting the optional target point position and all vertices, and the line connecting the vertices of different polygons are projected onto the area occupied by the warehouse to form a global environment model of the warehouse.

[0030] Optionally, the parsing of the storage in and out instruction of the material box robot to determine the current starting position and the current target point position of the material box robot includes:

[0031] Parsing the in-and-out instructions of the material box robot to obtain in-and-out configuration parameters;

[0032] According to the outbound parameters, it is determined whether it is an outbound instruction or an inbound instruction, and the current starting point position and the current target point position of the corresponding material box robot.

[0033] Optionally, the static path search for the in-and-out storage of the material box robot is performed according to the current starting point position, the current target point position and the global environment model to obtain a plurality of optional static paths to form an optional static path set, including:

[0034] According to the starting point position, the current target point position and the global environment model, a number of particles are constructed, and an initial fitness value and an initial inertia weight are assigned to each particle to obtain an initialized particle group, each particle representing an initialized static path;

[0035] The initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0036] Optionally, the initial fitness value represents the initial performance strength of the corresponding initialized static path, and the initial performance strength includes the length of the path, the initial number of turns made by the material box robot when moving along the initialized static path, and the average safety distance from obstacles.

[0037] Optionally, the initialized particle swarm is iterated to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including:

[0038] Based on the set path search mechanism, determine the self-cognition metric and the social cognition metric used when iterating the initialized particle group, the self-cognition metric is used to represent the initialization semantic distance between the current path length of the particle and its shortest path length, and the social cognition metric is used to represent the initialization semantic distance between the current path length of the particle and the global shortest path length;

[0039] Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0040] Optionally, the initialization particle swarm is iterated based on the self-cognition metric and the social cognition metric to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including:

[0041] Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated in the space where the initialized particle swarm is located, so as to update the particles in the initialized particle swarm in real time to obtain dynamic particles, so as to obtain a dynamic particle swarm, each dynamic particle having an updated fitness value and an updated inertia weight;

[0042] Based on the updated fitness value, some dynamic particles are screened out from the dynamic particle group to search for static paths for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0043] Optionally, the method further comprises:

[0044] According to the inertia weight update rule, the inertia weights of the particles in the initialized particle group are linearly updated so that the inertia weights decrease linearly with the increase of the number of iterations. The linear decrease makes it possible to search for a static path for entering and exiting the warehouse based on the nearest search position and search direction of the particles in the space where the initialized particle group is located.

[0045] Optionally, the solving of the optional static path set based on the set static path constraint to select an optional static path satisfying the set path constraint as the planned static path includes:

[0046] The static path constraint is set at a fitness threshold, and a global solution is performed with the updated fitness value of the updated particle corresponding to each static optional path in the optional static path set, so as to select the optional static path whose length, number of turns, and average safety distance from obstacles after the update path globally match the fitness threshold as the planned static path.

[0047] In the present invention, by reading the global environment model of the warehouse where the material box robot is located, the prior knowledge of the warehouse environment is fully utilized. The obstacle area and the free running area are defined in the global environment model, which enables path planning to be carried out under known environmental constraints, thereby avoiding the communication delay and perception error problems that may be encountered in dynamic path planning when perceiving the environment in real time. Secondly, the in-and-out instructions of the material box robot are parsed to determine the current starting position and target point position of the robot. This step enables path planning to be carried out based on specific task requirements, thereby improving the pertinence and practicality of path planning. Next, according to the starting point position, the target point position and the global environment model, a static path search for the in-and-out of the material box robot is performed. Several optional static paths are obtained through the search and constitute an optional static path set. This process makes full use of the information of the global environment model, reduces the amount of real-time calculation, and improves the efficiency of path planning. Finally, based on the set static path constraints, the optional static path set is solved, and the optional static path that meets the path constraints is selected as the static path for planning. This step ensures the feasibility and security of the selected path, and solves the path conflict problem when multiple tasks are concurrent. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Hereinafter, some specific embodiments of the present application will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:

[0049] Figure 1 A schematic flow chart of a static path planning method for a material box robot provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the embodiments of the present application should fall within the scope of protection of the embodiments of the present application.

[0051] Figure 1 A schematic diagram of a static path planning method for a container robot provided in an embodiment of the present application. Figure 1 As shown, it includes:

[0052] Reading a global environment model of the warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area;

[0053] Parsing the in-and-out instructions of the material box robot to determine the current starting position and the current target point position of the material box robot;

[0054] According to the starting point position, the current target point position and the global environment model, a static path for the material box robot to enter and exit the warehouse is searched to obtain a plurality of optional static paths to form an optional static path set;

[0055] Based on the set static path constraints, the optional static path set is solved to select an optional static path that meets the set path constraints and use it as the planned static path.

[0056] In the present invention, by reading the global environment model of the warehouse where the material box robot is located, the prior knowledge of the warehouse environment is fully utilized. The obstacle area and the free running area are defined in the global environment model, which enables path planning to be carried out under known environmental constraints, thereby avoiding the communication delay and perception error problems that may be encountered in dynamic path planning when perceiving the environment in real time. Secondly, the in-and-out instructions of the material box robot are parsed to determine the current starting position and target point position of the robot. This step enables path planning to be carried out based on specific task requirements, thereby improving the pertinence and practicality of path planning. Next, according to the starting point position, the target point position and the global environment model, a static path search for the in-and-out of the material box robot is performed. Several optional static paths are obtained through the search and constitute an optional static path set. This process makes full use of the information of the global environment model, reduces the amount of real-time calculation, and improves the efficiency of path planning. Finally, based on the set static path constraints, the optional static path set is solved, and the optional static path that meets the path constraints is selected as the static path for planning. This step ensures the feasibility and security of the selected path, and solves the path conflict problem when multiple tasks are concurrent.

[0057] Optionally, the inbound and outbound instructions include inbound instructions and / or outbound instructions, so as to complete an outbound or inbound task according to the task order, and multiple goods may be taken or unloaded at one time;

[0058] The warehousing instruction is used to schedule the material box robot to place the goods from the starting position to the warehousing workbench; correspondingly, the optional static path is: the route of the material box robot from its optional starting point position to the warehousing workbench, from the warehousing workbench to the first shelf, and from the first shelf to the next shelf until all the goods are placed;

[0059] The warehousing instruction is used to schedule the material box robot to pick up goods from the starting position to the warehousing workbench; correspondingly, the optional static path includes: the route of the material box robot from the starting position to the first shelf, the route from the first shelf to the next shelf until the goods required for the order are picked up, and the route from the last shelf to the outbound workbench.

[0060] Optionally, the reading of a global environment model of a warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area, further includes:

[0061] Obtaining a global environment model file of the warehouse where the material box robot is located, parsing the global environment model file, and determining the central coordinates of obstacles in the warehouse and the outlines of the obstacles;

[0062] A global environment model of the warehouse is constructed based on the central coordinates of the obstacle and the outline of the obstacle.

[0063] In the path planning of the bin robot, first obtaining and parsing the warehouse's global environment model file, and then building a detailed global environment model, this step brings significant technical benefits. The following is a detailed explanation of these benefits, as well as a simple exemplary code snippet.

[0064] To this end, the technical solution here has the following technical benefits:

[0065] (1) By parsing the global environment model file, the center coordinates and contours of obstacles in the warehouse can be accurately obtained, and then an accurate global environment model can be constructed. This enables the path planning algorithm to perform calculations based on accurate environmental information, thereby improving the accuracy of path planning.

[0066] At the same time, since the global environment model is pre-built, it can be directly used during path planning without the need for real-time perception of the environment, which greatly improves the efficiency of path planning.

[0067] (2) The global environment model contains detailed information about obstacles, which enables the path planning algorithm to fully consider the existence of obstacles and plan a path that avoids collisions with obstacles, thereby enhancing the safety of path planning.

[0068] (3) In the case of multiple tasks running concurrently, since the global environment model is shared, each task can perform path planning based on a unified environment model, which makes it easier to handle path conflicts and coordinate multiple tasks.

[0069] (4) The global environment model exists in the form of a file, which is easy to update and maintain. When the warehouse environment changes, you only need to update the global environment model file to make the path planning algorithm adapt to the new environment.

[0070] In addition, an exemplary code is provided to illustrate the implementation process of the above solution:

[0071]

[0072] In the above sample code, a load_global_environment_model function is first defined to load the global environment model file and parse it into an obstacle list. Then, a build_global_environment_model function is defined to build the global environment model based on the obstacle list. Finally, it shows how to use these two functions to load and build the global environment model.

[0073] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0074] Gridding the area occupied by the warehouse to obtain a plurality of grids to form a grid array;

[0075] Based on the central coordinates of the obstacle in the warehouse, matching the outline of the obstacle with the grid array to determine a mapping relationship between the obstacle and the grid array;

[0076] Based on the mapping relationship, the grid array is encoded to obtain obstacle position encoding and free area position encoding;

[0077] The obstacle position codes and free area position codes are projected onto the grid array to form a global environment model of the warehouse.

[0078] The technical solution of meshing the warehouse based on the center coordinates and outlines of obstacles and building a global environment model can bring many benefits. The following is a detailed explanation of these benefits and a simple exemplary code snippet.

[0079] To this end, a global environment model of the warehouse is constructed based on the central coordinates of the obstacles and the outlines of the obstacles, which has the following technical benefits:

[0080] (1) Meshing allows the warehouse environment to be represented with higher accuracy and resolution. Each grid can represent a small area in the warehouse, thus more accurately representing the location and contours of obstacles.

[0081] (2) By dividing the warehouse into grids and mapping obstacles to the grids, the complex path planning problem can be simplified to searching and calculating on the grid array. This greatly reduces the complexity of the calculation and improves the efficiency of path planning.

[0082] (3) When the warehouse environment changes, such as adding or removing obstacles, only the corresponding grids in the global environment model need to be updated without rebuilding the entire model. This makes it easier to dynamically update and maintain the environment model.

[0083] (4) The gridded global environment model can be easily visualized, which helps to better understand and debug the path planning algorithm. The location, outline, and free area of ​​obstacles can be intuitively seen, which can better optimize the path planning strategy.

[0084] In addition, an exemplary code is provided below to illustrate building a global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle:

[0085]

[0086]

[0087]

[0088] In the above sample code, a grid_warehouse function is first defined to create a grid array. Then, a map_obstacles_to_grid function is defined to map obstacles to the grid. Next, an encode_grid function is defined to encode the grid to obtain the obstacle position encoding and the free area position encoding. Finally, a build_global_environment_model function is defined to integrate these steps and build a global environment model.

[0089] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0090] Based on the center coordinates of the obstacle and the outline of the obstacle, defining a plurality of designated position coordinates that the material box robot will pass through when traveling in the warehouse to form a designated position coordinate sequence;

[0091] Determine the route of the material box robot from its starting position point coordinates to the nearest specified position coordinates, and the route from the nearest specified position coordinates to the next specified position coordinates, iterating in sequence to obtain a plurality of routes;

[0092] Map each route as an edge and mark obstacles on each edge;

[0093] All edges loaded with obstacle identifiers are spliced ​​together in the order of iteration to form a global environment model of the warehouse.

[0094] The global environment model of the warehouse is constructed based on the central coordinates and contours of the obstacles. By defining multiple specified position coordinate sequences that the box robot must pass through when moving in the warehouse, a series of routes are determined and mapped as edges, and finally the obstacle identifiers are loaded into the global environment model. This technical solution brings significant technical benefits.

[0095] To this end, constructing the global environment model of the warehouse based on the center coordinates of the obstacles and the outlines of the obstacles has the following technical benefits:

[0096] (1) By defining multiple specified position coordinates and forming a coordinate sequence, the container robot can flexibly select a travel route according to actual conditions and better adapt to the complexity and variability of the warehouse environment.

[0097] (2) Marking obstacles on each route can ensure that the container robot can detect and avoid obstacles in time during its movement, thereby avoiding collisions and damage and improving the safety of path planning.

[0098] (3) Representing the warehouse environment model as a series of edges with obstacle labels can simplify the calculation process of path planning. The container robot only needs to move in the order of the edges and handle obstacles accordingly when encountering them, without having to perform complex searches and calculations on the entire warehouse environment.

[0099] (4) When obstacles in the warehouse environment change, only the obstacle identification on the corresponding edge needs to be updated without rebuilding the entire global environment model. This makes the dynamic update and maintenance of path planning easier and more efficient.

[0100] (5) The global environment model is represented as a series of edges with obstacle labels, which can intuitively display the path of the container robot and the location of obstacles. This helps to better understand and optimize the path planning strategy and improve the interpretability of path planning.

[0101] In addition, an exemplary code is provided below to implement the construction of the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle:

[0102]

[0103]

[0104] In the above sample code, we first define the Node and Edge classes to represent nodes and edges. Then, we implement the build_global_environment_model function to build the global environment model. This function accepts a list of obstacles and a sequence of specified position coordinates as input, and returns a list of nodes and a list of edges. In the process of building edges, we check whether there are obstacles between each pair of nodes, and set the obstacle flags on the edges accordingly. Finally, we show how to use this function to build a global environment model.

[0105] Optionally, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes:

[0106] Generating polygons representing different obstacles based on the central coordinates of the obstacles and the outlines of the obstacles;

[0107] Determine the coordinates of each vertex of each polygon to establish a line connecting the optional starting point position and all vertices, to establish a line connecting the optional target point position and all vertices, and a line connecting vertices of different polygons;

[0108] The optional starting point position, the optional target point position, the polygon, the line connecting the optional starting point position and all vertices, the line connecting the optional target point position and all vertices, and the line connecting the vertices of different polygons are projected onto the area occupied by the warehouse to form a global environment model of the warehouse.

[0109] The global environment model of the warehouse is constructed based on the central coordinates of the obstacles and the outlines of the obstacles, which has the following technical advantages:

[0110] (1) By generating polygons representing different obstacles based on the center coordinates and contours of the obstacles, the shapes and locations of obstacles in the warehouse can be more accurately represented. This representation method can better reflect the actual space occupied by obstacles than simple points or lines, providing more accurate environmental information for path planning.

[0111] (2) Establishing the lines between the optional starting point and all polygon vertices, the lines between the optional target point and all polygon vertices, and the lines between different polygon vertices can ensure that the relative position relationship between all possible paths and obstacles is taken into account during path planning. This helps to avoid loopholes and collision risks in path planning.

[0112] (3) By projecting the optional starting point, target point, polygon and their connecting lines onto the area occupied by the warehouse, a complete global environment model can be formed. Path planning in this model can avoid searching and calculating irrelevant areas, thereby improving the efficiency of path planning.

[0113] (4) The global environment model provides a clear and structured representation of the environment for the path planning algorithm. The algorithm can perform path search and optimization based on this model without having to deal with complex obstacle shapes and position relationships. This simplifies the implementation process of the path planning algorithm.

[0114] (6) The global environment model can intuitively display the obstacle distribution and path planning results in the warehouse. This helps to visualize and debug the path planning algorithm, identify potential problems and optimize them.

[0115] In addition, the following provides exemplary code for constructing a global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle:

[0116]

[0117]

[0118] In the above sample code, a Polygon class is first defined to represent polygonal obstacles. Then, the generate_polygons function is implemented to generate a polygon list based on the obstacle list. Next, the build_global_environment_model function is implemented to build a global environment model, including establishing lines between the start point, the target point and all polygon vertices, and establishing lines between different polygon vertices. Finally, it shows how to use these functions to generate a global environment model. This model can provide a clear and structured environment representation for subsequent path planning algorithms.

[0119] Optionally, the parsing of the storage in and out instruction of the material box robot to determine the current starting position and the current target point position of the material box robot includes:

[0120] Parsing the in-and-out instructions of the material box robot to obtain in-and-out configuration parameters;

[0121] According to the outbound parameters, it is determined whether it is an outbound instruction or an inbound instruction, and the current starting point position and the current target point position of the corresponding material box robot.

[0122] To this end, the above-mentioned parsing of the in-and-out instructions of the material box robot to determine the current starting position and the current target point position of the material box robot has the following technical benefits:

[0123] (1) By parsing the in-and-out instructions of the bin robot, the in-and-out configuration parameters can be obtained, and then it can be accurately determined whether it is an out-and-out instruction or an in-and-out instruction, as well as the corresponding starting point position and target point position. This parsing method avoids ambiguity and misunderstanding of instructions and improves the accuracy and efficiency of instruction parsing.

[0124] (2) The system can flexibly handle different types of in-and-out instructions by parsing in-and-out instructions into configuration parameters. When adding a new in-and-out type or modifying an existing type, it only needs to adjust the configuration parameters without modifying the underlying code, which enhances the scalability of the system.

[0125] (3) After accurately parsing the in-and-out instructions, the clear starting point and target point positions can be obtained. This provides a clear goal and direction for subsequent path planning and navigation, helping the container robot to complete the in-and-out tasks quickly and accurately.

[0126] (4) When using the bin robot for warehousing and retrieval operations, users do not need to worry about complex command formats and parameter settings. The system will automatically parse the commands issued by the user and give corresponding feedback and execution results, which improves the user experience.

[0127] (5) Parsing the in-and-out instructions into configuration parameters makes troubleshooting and logging more convenient. When a fault occurs or historical operations need to be traced back, the problem can be quickly located and the cause analyzed by viewing the configuration parameters and log records.

[0128] In addition, the in-and-out instructions of the material box robot are parsed to determine the current starting position and the current target point position of the material box robot, and an exemplary code explanation is provided:

[0129]

[0130]

[0131] In the above sample code, an InventoryCommand class is defined to parse the in / out instructions. This class accepts a command string as input and parses it into configuration parameters through the parse_command method. Then, the class provides is_inbound and is_outbound methods to determine whether the command is in / out, as well as get_start_position and get_target_position methods to obtain the starting point and target point positions. Finally, it shows how to use this class to parse a specific in / out instruction and print out the parsing results.

[0132] Optionally, the static path search for the in-and-out storage of the material box robot is performed according to the current starting point position, the current target point position and the global environment model to obtain a plurality of optional static paths to form an optional static path set, including:

[0133] According to the starting point position, the current target point position and the global environment model, a number of particles are constructed, and an initial fitness value and an initial inertia weight are assigned to each particle to obtain an initialized particle group, each particle representing an initialized static path;

[0134] The initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0135] To this end, according to the current starting point position, the current target point position and the global environment model, a static path search for the material box robot in and out of the warehouse is performed to obtain several optional static paths to form an optional static path set, which has the following technical benefits:

[0136] (1) Using the particle swarm algorithm for path search can simultaneously consider multiple possible paths in the global environment model and find the optimal or suboptimal solution through iterative optimization. Compared with traditional single path search algorithms, such as Dijkstra or A* algorithms, this method can find a feasible path set faster and improve search efficiency.

[0137] (2) Each particle in the particle swarm algorithm represents a possible path. Through iterative updates, the algorithm can explore different path options, thereby increasing the diversity of paths. This is particularly important for a container robot, as it may need to find multiple alternative paths in a complex warehouse environment to deal with possible obstacles or traffic jams.

[0138] (3) Due to the iterative and optimization characteristics of the particle swarm algorithm, it can adapt to the dynamic changes of the warehouse environment. For example, if some areas in the warehouse suddenly become inaccessible, the algorithm can find new feasible paths by re-iteration without having to search again from scratch.

[0139] (4) Particle swarm optimization is a general optimization algorithm that can be easily integrated into the existing bin robot control system. In addition, the performance and results of the path search can be further optimized by adjusting the parameters of the algorithm (such as the number of particles, number of iterations, fitness function, etc.).

[0140] (5) By generating a set of optional static paths, the particle swarm algorithm provides a variety of possible path choices for the inbound and outbound tasks of the container robot. This allows the robot to make more intelligent decisions based on actual conditions (such as path length, travel time, obstacles, etc.) and choose the most appropriate path to perform the task.

[0141] In addition, according to the current starting point position, the current target point position and the global environment model, a static path search for the material box robot in and out of the warehouse is performed to obtain several optional static paths to form an optional static path set. The following exemplary code is provided for illustration:

[0142]

[0143]

[0144] In the above sample code, a Particle class is defined to represent particles in the particle swarm algorithm, and a ParticleSwarmOptimization class is defined to implement the particle swarm optimization algorithm. The algorithm searches for optional static paths for the container robot by iteratively updating the position and velocity of the particles, and finally returns a set of optional paths.

[0145] Optionally, the initial fitness value represents the initial performance strength of the corresponding initialized static path, and the initial performance strength includes the length of the path, the initial number of turns made by the material box robot when moving along the initialized static path, and the average safety distance from obstacles.

[0146] Combined with the technical solution, the initial fitness value represents the initial performance strength of the corresponding initialized static path, which has the following technical benefits:

[0147] (1) The initial fitness value comprehensively considers the length of the path, the number of turns the container robot makes when moving along the initialized static path, and the average safe distance from obstacles. This comprehensive evaluation method can more comprehensively reflect the performance of the path, rather than just judging based on a single indicator (such as path length).

[0148] (2) By taking the number of turns and the average safe distance to obstacles into account, the algorithm is able to prefer paths that are not only short but also smooth and safe. Such paths are more efficient for the container robot because they can reduce energy consumption and time delays when turning, while reducing the risk of collision with obstacles.

[0149] (3) The design of the initial fitness value makes the algorithm more flexible and can be adjusted according to different application scenarios and requirements. For example, in some cases, the safety distance may be more important than the path length. In this case, the weight in the fitness function can be adjusted to reflect this requirement.

[0150] (4) By assigning an initial fitness value to each particle, the algorithm can converge to the optimal or suboptimal solution more quickly during the iteration process. This is because particles are updated according to their fitness values, and a fitness value that comprehensively considers multiple performance indicators can more accurately guide the search direction of particles.

[0151] (5) In the warehousing and retrieval tasks of the bin robot, time is an important consideration. By designing the initial fitness value, the algorithm can generate a set of high-quality optional static paths in a shorter time, thereby improving the robot's decision-making efficiency. This means that the robot can select a suitable path and perform the task more quickly, thereby improving overall work efficiency.

[0152] Optionally, the initialized particle swarm is iterated to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including:

[0153] Based on the set path search mechanism, determine the self-cognition metric and the social cognition metric used when iterating the initialized particle group, the self-cognition metric is used to represent the initialization semantic distance between the current path length of the particle and its shortest path length, and the social cognition metric is used to represent the initialization semantic distance between the current path length of the particle and the global shortest path length;

[0154] Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0155] To this end, the initialization of the particle swarm is iterated, and self-cognition metrics and social cognition metrics are introduced to guide the search process, which has the following technical benefits:

[0156] (1) The introduction of self-cognition metric and social cognition metric enables particles to adjust their positions and speeds more specifically during the iteration process to search for a better path. The self-cognition metric helps particles understand the improvement of their own paths, while the social cognition metric lets particles know the gap with the global optimal path, thereby jointly guiding the search direction of particles.

[0157] (2) Guided by these two metrics, the particle swarm algorithm can converge to the optimal or suboptimal solution more quickly. During the iteration process, particles are more inclined to explore areas that may produce better paths, thereby reducing invalid searches and improving overall search efficiency.

[0158] (3) Although the self-cognition metric and social cognition metric will guide the particles to converge to a better path, due to the parallelism and randomness of the particle swarm algorithm itself, it can still maintain a certain path diversity. This means that the algorithm can explore and return multiple different optional paths, providing more options for the in-and-out storage tasks of the bin robot.

[0159] (4) This design makes the particle swarm algorithm more adaptable to complex warehouse environments. Whether facing static obstacles or dynamically changing traffic conditions, the algorithm can adapt to different environmental conditions by adjusting the weights of self-cognition metrics and social cognition metrics to find a suitable path.

[0160] (5) By introducing these two metrics, the particle swarm algorithm not only considers the length of the path during the search process, but also implicitly considers the smoothness and safety of the path. This enables the algorithm to make more intelligent decisions when selecting paths for the container robot, choosing those paths that are short, smooth, and safe.

[0161] In addition, the following is an example code to illustrate how to introduce self-cognition metrics and social cognition metrics into the particle swarm algorithm:

[0162]

[0163]

[0164] In this example, the update_velocity method is modified to include a self-cognition part and a social-cognition part, which are calculated by self_cognition and social_cognition respectively. These two parts together guide the particle to adjust its speed and position during the iteration process to search for a better path.

[0165] Optionally, the initialization particle swarm is iterated based on the self-cognition metric and the social cognition metric to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including:

[0166] Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated in the space where the initialized particle swarm is located, so as to update the particles in the initialized particle swarm in real time to obtain dynamic particles, so as to obtain a dynamic particle swarm, each dynamic particle having an updated fitness value and an updated inertia weight;

[0167] Based on the updated fitness value, some dynamic particles are screened out from the dynamic particle group to search for static paths for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

[0168] To this end, the initialization particle swarm is iterated based on the self-cognition metric and the social cognition metric, and the particles are updated in real time to obtain a dynamic particle swarm, and then some dynamic particles are selected from them to search for static paths, which has the following technical benefits:

[0169] (1) Real-time updating of particles to obtain a dynamic particle swarm enables the algorithm to dynamically adapt to changes in the warehouse environment. Whether it is the change in the position of a static obstacle or the appearance of a dynamic obstacle, the algorithm can adapt to these changes by updating the position and velocity of the particles, thereby finding a new feasible path.

[0170] (2) With the guidance of self-cognition and social cognition metrics, dynamic particle swarms can converge to the optimal or suboptimal solution more quickly. During the iteration process, particles are more inclined to explore areas that may produce better paths, reducing invalid searches and thus improving search efficiency.

[0171] (3) Due to the parallelism and randomness of the particle swarm algorithm itself, the dynamic particle swarm can still maintain a certain path diversity even with real-time updates. This means that the algorithm can explore and return multiple different optional paths, providing more options for the bin robot's warehousing and outbound tasks.

[0172] (4) The design of the dynamic particle swarm is updated in real time, making the algorithm less sensitive to the selection of the initial particle swarm. Even if the quality of the initial particle swarm is not high, the algorithm can gradually find a better path through iteration and updating, thereby enhancing the robustness of the algorithm.

[0173] (5) By introducing self-cognition metrics and social cognition metrics, combined with a real-time update mechanism, the particle swarm algorithm can make more intelligent decisions when selecting paths for the container robot. The algorithm not only considers the length of the path, but also implicitly considers the smoothness and safety of the path, thereby selecting a more appropriate path.

[0174] In addition, the following is an exemplary code to illustrate how to iterate the initialized particle swarm based on the self-cognition metric and the social cognition metric and obtain a dynamic particle swarm:

[0175]

[0176] In this example, the Particle class adds an update method to update the particle's speed and position and calculate the new fitness value and inertia weight. The optimize method of the ParticleSwarmOptimization class is responsible for iterative optimization and selects some particles from the dynamic particle swarm based on the updated fitness value after each iteration to search for static paths for the bin robot to enter and exit the warehouse. Finally, the positions of particles with fitness values ​​greater than a certain threshold are returned as a set of optional static paths.

[0177] Optionally, the method further comprises:

[0178] According to the inertia weight update rule, the inertia weights of the particles in the initialized particle group are linearly updated so that the inertia weights decrease linearly with the increase of the number of iterations. The linear decrease makes it possible to search for a static path for entering and exiting the warehouse based on the nearest search position and search direction of the particles in the space where the initialized particle group is located.

[0179] Incorporating the linear update strategy of inertia weight into the particle swarm optimization algorithm brings significant technical advantages to the static path search of the bin robot in and out of the warehouse:

[0180] To this end, the scheme based on the above-mentioned inertia weight update rule has the following technical advantages:

[0181] (1) As the number of iterations increases, the inertia weight decreases linearly, which means that when the particle updates its velocity and position, its reliance on the current position and velocity gradually decreases, while its reliance on self-cognition and social cognition increases relatively. Such adjustments allow the algorithm to focus more on local search in the later stages of iterations, which helps to find a better path.

[0182] (2) The linear decrease strategy of inertia weight establishes a dynamic balance between the global search and local search of the algorithm. In the early stage of iteration, a larger inertia weight helps particles explore a wider space, while as the weight decreases, particles can search the potential optimal path area more finely.

[0183] (3) By adjusting the inertia weight, the algorithm can converge to the optimal or suboptimal solution more effectively. The linear reduction strategy ensures that the particles maintain sufficient exploratory power during the iteration process and gradually focus on the most promising search area.

[0184] (4) This strategy reduces unnecessary extensive searches, allowing the algorithm to locate a feasible path more quickly. This is particularly important for bin robot warehousing tasks that require high real-time performance.

[0185] (5) The linear decreasing inertia weight strategy enables the algorithm to flexibly adapt to different warehouse environments and task requirements. By adjusting the decreasing rate and initial weight, the algorithm can be optimized for different scenarios.

[0186] In addition, a code example is provided below to show how to implement the linear update of inertia weight in the particle swarm optimization algorithm:

[0187] classParticleSwarmOptimization:

[0188] def__init__(self,num_particles,num_iterations,inertia_weight_start,inertia_weight_end):

[0189] self.inertia_weight_start=inertia_weight_start#initial inertia weight

[0190] self.inertia_weight_end=inertia_weight_end#Final inertia weight

[0191] self.inertia_weight_decrement = (self.inertia_weight_start - self.inertia_weight_end) / num_iterations # inertia weight decrease

[0192] def optimize(self):

[0193] for iin range(self.num_iterations):

[0194] current_inertia_weight = self.inertia_weight_start-i*self.inertia_weight_decrement # Calculate the current inertia weight

[0195] for particle in self.particles:

[0196] particle.update(self.global_best,current_inertia_weight,self.c1,self.c2)#Update particles using current inertia weight

[0197] In this example, the constructor of the ParticleSwarmOptimization class adds two new parameters, inertia_weight_start and inertia_weight_end, which represent the starting and ending values ​​of the inertia weight. inertia_weight_decrement is used to calculate the decrement of the inertia weight at each iteration. In the optimize method, each iteration calculates the current inertia weight based on the current number of iterations and uses this weight to update the particles. This strategy helps the algorithm gradually focus on more promising areas during the search process, thereby improving search efficiency and path quality.

[0198] Optionally, the solving of the optional static path set based on the set static path constraint to select an optional static path satisfying the set path constraint as the planned static path includes:

[0199] The static path constraint is set at a fitness threshold, and a global solution is performed with the updated fitness value of the updated particle corresponding to each static optional path in the optional static path set, so as to select the optional static path whose length, number of turns, and average safety distance from obstacles after the update path globally match the fitness threshold as the planned static path.

[0200] To this end, by solving the optional static path set based on the set static path constraints, an optional static path that satisfies the path constraints is selected as the planned static path, which has the following technical benefits:

[0201] (1) By setting the fitness threshold and considering multiple dimensions such as the length of the path, the number of turns, and the average safe distance from obstacles, the selected path can be ensured to be feasible in practical applications. Such a path not only meets the robot's movement capabilities, but also effectively avoids collisions with obstacles in the warehouse.

[0202] (2) The global solution process comprehensively considers multiple path evaluation indicators, so that a path with excellent performance in multiple aspects can be selected. Such a path is often smoother, more efficient, and has better security.

[0203] (3) By setting different fitness thresholds, the algorithm’s path selection criteria can be flexibly adjusted. This enables the algorithm to adapt to different warehouse environments and task requirements, and provide more personalized path planning solutions for the inbound and outbound tasks of the container robot.

[0204] (4) Comprehensively considering multiple path constraints and evaluation indicators for global solution reflects the intelligence of the algorithm in path selection. Such a decision-making process is more comprehensive and detailed, and can ensure that the selected path meets the optimal or suboptimal standards in multiple aspects.

[0205] (5) Since the selected path has satisfied all static path constraints, no additional path verification and optimization work is required in the subsequent processing flow. This greatly simplifies the processing flow and improves the overall efficiency of the algorithm.

[0206] In addition, the following is an example code to illustrate how to solve a set of optional static paths based on the set static path constraints:

[0207]

[0208]

[0209] In this example, the PathConstraints class is used to define static path constraints, including the maximum length of the path, the maximum number of turns, and the minimum safe distance from obstacles. The evaluate_path function is used to evaluate whether a path meets these constraints. The select_optimal_path function traverses the set of optional static paths and selects the path that meets the constraints as the static path for planning. This design enables the algorithm to more intelligently and flexibly select high-quality paths that meet multiple constraints.

[0210] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A static path planning method for a material box robot entering and exiting a warehouse, characterized in that: include: Reading a global environment model of the warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area; Parsing the in-and-out instructions of the material box robot to determine the current starting position and the current target point position of the material box robot; According to the starting point position, the current target point position and the global environment model, a static path for the material box robot to enter and exit the warehouse is searched to obtain a plurality of optional static paths to form an optional static path set; Based on the set static path constraints, solving the optional static path set to select an optional static path that satisfies the set path constraints as the planned static path; The inbound and outbound instructions include inbound instructions and / or outbound instructions, so as to complete an outbound or inbound task according to the task order, and multiple goods can be taken or unloaded at one time; The warehousing instruction is used to schedule the material box robot to place the goods from the starting position to the warehousing workbench; correspondingly, the optional static path is: the route of the material box robot from its optional starting point position to the warehousing workbench, from the warehousing workbench to the first shelf, and from the first shelf to the next shelf until all the goods are placed; The warehousing instruction is used to schedule the material box robot to pick up goods from the starting position to the warehousing workbench; correspondingly, the optional static path includes: the route of the material box robot from the starting position to the first shelf, the route from the first shelf to the next shelf until the goods required for the order are picked up, and the route from the last shelf to the outbound workbench.

2. The static path planning method of a material box robot according to claim 1, characterized in that: The reading of the global environment model of the warehouse where the material box robot is located, wherein the global environment model defines an obstacle area and a free running area, further includes: Obtaining a global environment model file of the warehouse where the material box robot is located, parsing the global environment model file, and determining the central coordinates of obstacles in the warehouse and the outlines of the obstacles; Constructing a global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle; Wherein, constructing the global environment model of the warehouse based on the central coordinates of the obstacle and the outline of the obstacle includes: Gridding the area occupied by the warehouse to obtain a plurality of grids to form a grid array; Based on the central coordinates of the obstacle in the warehouse, matching the outline of the obstacle with the grid array to determine a mapping relationship between the obstacle and the grid array; Based on the mapping relationship, the grid array is encoded to obtain obstacle position encoding and free area position encoding; The obstacle position codes and free area position codes are projected onto the grid array to form a global environment model of the warehouse.

3. The static path planning method of a material box robot according to claim 2, characterized in that: The global environment model of the warehouse is constructed based on the central coordinates of the obstacle and the outline of the obstacle, including: Based on the center coordinates of the obstacle and the outline of the obstacle, defining a plurality of designated position coordinates that the material box robot will pass through when traveling in the warehouse to form a designated position coordinate sequence; Determine the route of the material box robot from its starting position point coordinates to the nearest specified position coordinates, and the route from the nearest specified position coordinates to the next specified position coordinates, iterating in sequence to obtain a plurality of routes; Map each route as an edge and mark obstacles on each edge; All edges loaded with obstacle identifiers are spliced ​​together in the order of iteration to form a global environment model of the warehouse.

4. The static path planning method of a material box robot according to claim 2, characterized in that: The global environment model of the warehouse is constructed based on the central coordinates of the obstacle and the outline of the obstacle, including: Generating polygons representing different obstacles based on the central coordinates of the obstacles and the outlines of the obstacles; Determine the coordinates of each vertex of each polygon to establish a line connecting the optional starting point position and all vertices, to establish a line connecting the optional target point position and all vertices, and a line connecting vertices of different polygons; The optional starting point position, the optional target point position, the polygon, the line connecting the optional starting point position and all vertices, the line connecting the optional target point position and all vertices, and the line connecting the vertices of different polygons are projected onto the area occupied by the warehouse to form a global environment model of the warehouse.

5. The static path planning method of a material box robot according to claim 1, characterized in that: The parsing of the in-and-out instructions of the material box robot to determine the current starting position and the current target point position of the material box robot includes: Parsing the in-and-out instructions of the material box robot to obtain in-and-out configuration parameters; According to the outbound parameters, it is determined whether it is an outbound instruction or an inbound instruction, and the current starting point position and the current target point position of the corresponding material box robot.

6. The static path planning method of a material box robot according to claim 1, characterized in that: According to the current starting point position, the current target point position and the global environment model, a static path search for the material box robot entering and exiting the warehouse is performed to obtain a plurality of optional static paths to form an optional static path set, including: According to the starting point position, the current target point position and the global environment model, a number of particles are constructed, and an initial fitness value and an initial inertia weight are assigned to each particle to obtain an initialized particle group, each particle representing an initialized static path; The initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

7. The static path planning method of a material box robot according to claim 8, characterized in that: The initialized particle swarm is iterated to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including: Based on the set path search mechanism, determine the self-cognition metric and the social cognition metric used when iterating the initialized particle group, the self-cognition metric is used to represent the initialization semantic distance between the current path length of the particle and its shortest path length, and the social cognition metric is used to represent the initialization semantic distance between the current path length of the particle and the global shortest path length; Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated to search for a static path for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

8. The static path planning method of a material box robot according to claim 7, characterized in that: Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated to search for a static path for the box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set, including: Based on the self-cognition metric and the social cognition metric, the initialized particle swarm is iterated in the space where the initialized particle swarm is located, so as to update the particles in the initialized particle swarm in real time to obtain dynamic particles, so as to obtain a dynamic particle swarm, each dynamic particle having an updated fitness value and an updated inertia weight; Based on the updated fitness value, some dynamic particles are screened out from the dynamic particle group to search for static paths for the material box robot to enter and exit the warehouse, and a plurality of optional static paths are obtained to form an optional static path set.

9. The static path planning method of a material box robot according to claim 1, characterized in that: The method further comprises: According to the inertia weight update rule, the inertia weights of the particles in the initialized particle group are linearly updated so that the inertia weights decrease linearly with the increase of the number of iterations. The linear decrease makes it possible to search for a static path for entering and exiting the warehouse based on the nearest search position and search direction of the particles in the space where the initialized particle group is located.

10. The static path planning method of a material box robot according to claim 1, characterized in that: The step of solving the set of optional static paths based on the set static path constraint to select an optional static path satisfying the set path constraint as the planned static path includes: The static path constraint is set at a fitness threshold, and a global solution is performed with the updated fitness value of the updated particle corresponding to each static optional path in the optional static path set, so as to select the optional static path whose length, number of turns, and average safety distance from obstacles after the update path globally match the fitness threshold as the planned static path.

Citation Information

Patent Citations

  • Robot whole-situation path planning method facing uncertain environment of mixed terrain and region

    CN102854880A

  • Path planning method of multi-robot system

    CN111912407A

  • Path planning method for mobile robot between ridges

    CN117193326A

  • Path planning method and system based on visibility graph and D*Lite algorithm

    CN117419738A

  • Multi-frame fusion method and apparatus for grounding contour line of stationary obstacle, and medium

    WO2022252380A1