A global path planning method for mobile robots
By extracting and storing backbone paths and using hybrid A* algorithms and distance measurement methods, the problem of low efficiency in global path planning of mobile robots is solved, and fast and efficient path planning is achieved, suitable for autonomous vehicles and service robots.
Patent Information
- Application Number
- CN202111419929.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-26
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2041-11-26
AI Technical Summary
The existing global path planning method for mobile robots is inefficient in dealing with the robot's own kinematics and environmental obstacle constraints, resulting in slow search speed and low path quality, hindering the large-scale application of robots.
A memory-based global path planning method is proposed. By extracting backbone paths and storing them in the database, repetitive calculations are reduced, and local paths are quickly planned and spliced into complete paths using mixed A* algorithm and Reeds-Shepp curve length or Euro-style distance measurements. .
It significantly improves the efficiency of path planning and reduces the search space. Especially in the large environmental area, the backbone path network can further reduce the amount of repeated planning and calculations, and improves application efficiency.
Smart Images

Figure CN114253258B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of mobile robots, and in particular relates to a global path planning method for a mobile robot. Background Art
[0002] In recent years, with breakthroughs in key technologies such as positioning and perception, intelligent mobile robots represented by unmanned delivery vehicles and automated guided vehicles (AGVs) have been rapidly applied in fields such as autonomous driving and unmanned logistics, gradually improving people's production and life.
[0003] Path planning is a key technology to achieve autonomous mobility of robots. The current global path planning technology still has problems such as slow search speed and low path quality when dealing with complex constraints such as the robot's own kinematics and environmental obstacles, which hinders the large-scale application of mobile robots. Summary of the invention
[0004] Aiming at the problem of low efficiency of existing global path planning methods for mobile robots, the present invention proposes a global path planning method for mobile robots, which can reduce repeated calculations by repeatedly utilizing memory paths, thereby greatly improving planning efficiency.
[0005] Embodiments of the present invention may be applied to intelligent mobile robots including autonomous driving vehicles, automated guided vehicles (AGVs), and service robots.
[0006] The technical solution proposed by the present invention comprises the following steps:
[0007] A global path planning method for a mobile robot, the planning method comprising the following steps:
[0008] Step 1: Plan a complete feasible path for the mobile robot based on the starting point, target point, and obstacle distribution;
[0009] Step 2, from the feasible path, remove the points whose distances from the starting point and the target point are less than a set length d0, to form a backbone path including a new starting point and a new end point, and store the starting point and the end point of the backbone path as indexes in the database;
[0010] Step 3: When planning again, search the database according to the starting point and target point of the new task and extract the corresponding backbone path;
[0011] Step 4: Use the local path planning method to connect the new starting point and target point to the backbone path to obtain the final feasible path.
[0012] Furthermore, the set length d0 is determined according to the minimum turning radius of the mobile robot and is specifically 2R min <d0<4R min , Rmin is the minimum turning radius.
[0013] Furthermore, in step 1, a hybrid A* algorithm is used to plan a complete feasible path for the mobile robot.
[0014] Furthermore, the backbone path is obtained according to the following method:
[0015] The sum of the distance between the starting point of the new task and the starting point of the backbone path and the distance between the target point of the new task and the end point of the backbone path is the smallest, and the backbone path is the backbone path corresponding to the new task.
[0016] As a preferred embodiment of the present application, in step 3, if the mobile robot is a car-like robot, the length of the Reeds-Shepp curve is used as the distance metric;
[0017] If the robot is an omnidirectional mobile robot, the Euclidean distance metric is used.
[0018] As a preferred implementation scheme of the present application, a hybrid A* algorithm is used to connect the starting point of the new task and the starting point of the backbone path; the hybrid A* algorithm is used to connect the end point of the backbone path and the target point of the new task; and then the three paths are spliced into a complete path as a global path that meets the obstacle avoidance and mobile robot motion constraints.
[0019] Compared with the prior art, the technical solution of the present invention has the following advantages:
[0020] The backbone path is a feasible path segment that satisfies obstacle avoidance and motion constraints. By using the memorized backbone path, the robot's search space will be significantly reduced, and the global path planning of the AGV will be converted into local path planning within the vicinity of the robot. By reusing the memorized path, repeated calculations can be reduced, thereby greatly improving planning efficiency. Moreover, the larger the environmental area, the longer the backbone path, and the more obvious the efficiency improvement. After the robot's operating point covers a certain area of the environment, the backbone path will form a backbone path network, which can be used to further reduce the amount of repeated planning calculations. Therefore, this invention has greater room for improvement. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0022] Figure 1 It is a working scenario diagram for global path planning of a mobile robot;
[0023] Figure 2 It is a schematic diagram of the backbone path extracted from the feasible path;
[0024] Figure 3 This is a schematic diagram of the global path planning method for a mobile robot provided by this application. DETAILED DESCRIPTION
[0025] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below by taking the automatic handling task of an indoor forklift robot as an example. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0026] The combination of robotics and bionics is one of the future development trends of robots. After millions of years of evolution, animals can quickly generate high-quality motion paths in complex natural environments. By drawing on the efficient motion planning mechanism of animals and integrating it with existing methods, the path planning efficiency of robots can be significantly improved.
[0027] According to the study of the hippocampus, the neural basis of mammalian navigation, animals can store paths in memory when planning paths, and recall their historical paths when searching for paths later, that is, path planning does not start completely from scratch. Inspired by this idea, the present invention proposes a memory-based global path planning method for robots.
[0028] The car-like robot model is used to model the kinematics of the indoor unmanned forklift. The car-like robot belongs to a non-holonomic constraint system with a minimum turning radius R min .
[0029] The environment map is represented by polygonal geometry, and the polygon is used to represent the plane projection of the minimum bounding box of the forklift robot, such as Figures 1 to 3 As shown, in this application, a rectangle is used as the external shape of the bounding box, and the collision detection between the robot and the environment adopts the AABB Axially Aligned Bounding Box bounding box intersection detection algorithm.
[0030] Figure 1A working scenario provided for an embodiment of the present invention, in which a straight line is an obstacle, a rectangle is a robot bounding box, and no collision is indicated if the bounding box does not intersect with the obstacle; between the starting posture and the target posture shown in the figure, a feasible path is obtained by planning using the hybrid A* algorithm, that is, the medial axis curve of the bounding box movement. The robot bounding box, i.e., a square box, at discrete path points is also drawn on the path. Since the robot has kinematic constraints, the hybrid A* path planning algorithm is used as the backbone path construction algorithm. According to the given starting and ending posture coordinates, the hybrid A* algorithm is called to generate a feasible path that avoids intersections with obstacles and satisfies the robot motion constraints.
[0031] Extract the backbone path from the feasible path. Eliminate the feasible path points within a distance of d0 from the starting point, and also eliminate the feasible path points within a distance of d0 from the end point. Record the remaining feasible path points as the backbone path m , the new backbone path has a new start and end point. The choice of the distance threshold d0 depends on the robot’s minimum turning radius R min In this embodiment, d0=2R is selected min , take the point on the backbone path closest to the original starting point as the new starting point p1, and take the point closest to the original end point as the new end point p2, see Figure 2 Each waypoint is a three-dimensional vector consisting of the robot's xy position and yaw angle θ.
[0032] Construct a new starting point and end point pair as index index = (p1, p2), associate the index with the backbone path and store it in the database. m}. If there is a new transport task, Figure 3 The starting posture 3 is denoted as p3, and the ending posture 4 is denoted as p4. Then the index of the backbone path is retrieved according to the distance, and the index with the smallest distance is found. The sum of the distance between the starting point of the new task and the starting point of the backbone path plus the distance between the target point of the new task and the end point of the backbone path is taken as the minimum distance, which is denoted as distance d. For car-like robots, the calculation formula for distance d is d=ReedsShepp(p1, p3)+ReedsShepp(p2, p4). The ReedsShepp(s, t) function represents the length of the Reeds-Shepp curve between posture s and posture t, which is the sum of posture (s, t) and the minimum turning radius R min For an omnidirectional mobile robot, define d = L2(p1, p3) + L2(p2, p4), where the L2 function represents the L2 norm, that is, the sum of the squares of each element, and then take the square root. The mod is the modulo function, where p x ,p y is the xy position coordinate of pose p, pθ ,q θ is the posture coordinate of p; select the backbone path corresponding to the index with the smallest d.
[0033] Use the Reeds-Shepp curve or the hybrid A* algorithm to construct a local path between p3 and p1 (denoted as path s ), use the same method to construct the local path between p4 and p2 (referred to as path t ).
[0034] The two local paths path s 、path t With backbone path m Splice into a complete path path Join =(path s ,path m ,path t ), the path is directly output to the robot's path tracking module as the result of the path planning module, see Figure 3 .
[0035] The backbone path is a feasible path segment that satisfies obstacle avoidance and motion constraints. By using the memorized backbone path, the robot's search space will be significantly reduced, and the global path planning of the AGV will be converted into local path planning within the vicinity of the robot. By reusing the memorized path, repeated calculations can be reduced, thereby greatly improving planning efficiency. Moreover, the larger the environmental area, the longer the backbone path, and the more obvious the efficiency improvement. After the robot's operating point covers a certain area of the environment, the backbone path will form a backbone path network, which can be used to further reduce the amount of repeated planning calculations. Therefore, this invention has greater room for improvement.
[0036] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and adjustments can be made without departing from the principle of the present invention. These improvements and adjustments should also be regarded as the scope of protection of the present invention.
Claims
1. A global path planning method for a mobile robot, characterized in that: The planning method comprises the following steps: Step 1: Plan a complete feasible path for the mobile robot based on the starting point, target point, and obstacle distribution; Step 2: Remove from the feasible paths the paths whose distances from the starting point to the target point are less than a set length. d 0, forming a backbone path including a new starting point and a new end point, and storing the starting point and the end point of the backbone path in the database as indexes; Step 3: When planning again, search the database according to the starting point and target point of the new task and extract the corresponding nearest backbone path; Step 4: Use the local path planning method to connect the new starting point and target point to the nearest backbone path to obtain the final feasible path.
2. A mobile robot global path planning method according to claim 1, characterized in that: The set length d 0 According to the minimum turning radius of the mobile robot R min OK, specifically 2 R min < d 0<4 R min .
3. A mobile robot global path planning method according to claim 1, characterized in that: In step 1, a hybrid A* algorithm is used to plan a complete feasible path for the mobile robot.
4. A mobile robot global path planning method according to claim 1, characterized in that: In step 3, the corresponding nearest backbone path is obtained according to the following method: The sum of the distance between the starting point of the new task and the starting point of the backbone path and the distance between the target point of the new task and the end point of the backbone path is the smallest, and the backbone path is the backbone path corresponding to the new task.
5. A mobile robot global path planning method according to claim 4, characterized in that: In step 3, if the mobile robot is a car-like robot, the length of the Reeds-Shepp curve is used as the distance metric; If the robot is an omnidirectional mobile robot, the Euclidean distance metric is used.
6. A mobile robot global path planning method according to claim 4, characterized in that: In step 4, the hybrid A* algorithm is used to connect the starting point of the new task and the starting point of the backbone path; the hybrid A* algorithm is used to connect the end point of the backbone path and the target point of the new task; and then the three paths are spliced into a complete path as a global path that meets the obstacle avoidance and mobile robot motion constraints.
