Path planning and obstacle avoidance method and system for multiple four-way shuttle vehicles for single target task

Abstract

The invention discloses a single-target task multi-four-way shuttle vehicle path planning and obstacle avoidance method and system, and the method comprises the steps: obtaining the basic information of a warehousing system, and building a four-way shuttle vehicle warehousing system grid map; establishing a task model with the goal of minimizing the total time required by all the four-way shuttle vehicles to complete the single target task; establishing a CBS-ET algorithm solving model, solving the task model, and performing path planning; the CBS-ET algorithm solution model adopts a layered search strategy, an upper layer algorithm plans a plurality of initial path sets of the four-way shuttle vehicles in a four-way shuttle vehicle storage system grid map, and conflict detection is carried out on the initial path sets; when a path conflict occurs, performing local planning on the path by using a lower-layer algorithm, and changing and optimizing the path; continuously iterating the upper-layer algorithm and the lower-layer algorithm, and gradually finding a group of conflict-free optimal paths through circulation of conflict detection and path re-planning; and the multiple four-way shuttle vehicles complete a single-target task according to the optimal path obtained through solving.

Description

Technical Field

[0001] The present invention belongs to the technical field of path planning, and in particular to a single-target task multi-four-way shuttle path planning and obstacle avoidance method and system. Background Art

[0002] As an emerging intelligent warehousing solution, the four-way shuttle storage system uses a four-way movable shuttle as the main handling equipment, providing more flexible storage and access capabilities. However, this technology also brings its own challenges. With the increase in the number of equipment and the complexity of the collaborative operation of different equipment, the task scheduling of the four-way shuttle storage system has become more cumbersome, and multiple four-way shuttles may conflict and lock when operating in parallel on the same floor, resulting in low overall operating efficiency and poor stability. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a single-target task multi-four-way shuttle path planning and obstacle avoidance method and system, which can minimize the number of turns and task completion time by optimizing the driving path of the four-way shuttle, effectively avoid potential multi-vehicle conflicts, and further improve the overall operation efficiency and stability.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A single-target task multi-four-way shuttle path planning and obstacle avoidance method includes the following steps:

[0006] Obtain basic information of the storage system and establish a grid map of the four-way shuttle storage system;

[0007] A task model is established with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task;

[0008] A CBS-ET algorithm solution model is established to solve the task model and perform path planning. The CBS-ET algorithm solution model adopts a hierarchical search strategy. The upper-level algorithm plans multiple initial path sets of four-way shuttles in the grid map of the four-way shuttle warehouse system and performs conflict detection on the initial path sets. When a path conflict occurs, the lower-level algorithm is used to perform local planning on the path and change the optimized path. The upper-level algorithm and the lower-level algorithm are continuously iterated, and a set of conflict-free optimal paths are gradually found through a cycle of conflict detection and path replanning.

[0009] Multiple four-way shuttles avoid obstacles according to the optimal path and complete single-target tasks.

[0010] Preferably, the basic information of the warehousing system includes information on the number of main lanes, information on the number of sub-lanes, information on the layout of cargo channels, and information on the number of shuttle vehicles.

[0011] Preferably, the task model is:

[0012]

[0013] in: is the time when Rj departs from the starting point, is the time when Rj arrives at the destination; n is the number of four-way shuttles assigned with single-target tasks, and its number belongs to J={1,2,...,n}, Rj(j∈J) represents the jth four-way shuttle; C Rj ={s Rj ,e Rj} is a single-objective task, s Rj ∈G and e Rj ∈G are single-objective tasks C Rj The starting and ending positions of Rj are the starting and ending points of the path Rj; cost(Rj) represents the cost of Rj performing the transport task C. Rj The objective function of the task model is to minimize the total time required for all four-way shuttles to complete a single-target task.

[0014] Preferably, after establishing the task model, the task model is constrained;

[0015] The constraints include:

[0016] w Rj (t)∈G;

[0017] z(w Rj (t))=0,z∈G;

[0018]

[0019] w Rj (t)≠w Rj′ (t),j′∈J,j≠j′;

[0020] w Rj (t)≠w Rj′ (t+1)∧w Rj (t+1)≠w Rj′ (t);

[0021] Among them, w Rj (t)∈G represents the path of Rj in the grid map of the four-way shuttle warehouse system; z(w Rj (t)) = 0, z∈G means that the path of Rj does not pass through the obstacle point, and z is the obstacle point in the grid map of the four-way shuttle warehouse system; Indicates the starting and ending points of the Rj path corresponding to the starting and ending positions of the task; Indicates that in each time unit Rj can only move to the adjacent position or wait in place, and Respectively represent the position of Rj at time t and t+1; w Rj (t)≠w Rj′ (t), j′∈J, j≠j′ means that two four-way shuttles cannot be at the same position on the map at the same time; w Rj (t)≠w Rj′ (t+1)∧w Rj (t+1)≠w Rj′ (t) indicates that the four-way shuttle cannot interchange positions while moving.

[0022] Preferably, a CBS-ET algorithm solution model is established to solve the task model and perform path planning; specifically, the following steps are included:

[0023] Step 1: Establish a storage environment map, and initialize the number, starting and ending positions, and motion parameters of the four-way shuttles;

[0024] Step 2, using the upper layer algorithm to plan a set of initial paths for multiple four-way shuttles in the four-way shuttle warehouse system grid map;

[0025] Step 3, perform conflict detection on the initial path set in step 2; when a path conflict occurs, execute step 4, and if there is no path conflict, directly output the conflict-free optimal path;

[0026] Step 4: Add constraints to the multiple four-way shuttles to ensure that one of the four-way shuttles avoids conflicts and replans the path;

[0027] Step 5: Use the lower-level algorithm to perform local planning on the path and change the optimized path;

[0028] Step 6: The optimized path returns to step 3 to re-check conflicts, and the above steps are repeated until there are no more path conflicts, and the conflict-free optimal path is output.

[0029] Furthermore, in step 4, if there is a conflict that Rj, Rj′ (j, j′∈J) occupy a node s at the same time t, it is recorded as (Rj, Rj′, s, t); if there is a conflict that Rj, Rj′ travel in opposite directions through an edge at the same time <v j ,v j′ >, then it is recorded as (Rj,Rj′,v j ,v j′ ,t); if Rj,Rj′ has a node conflict c jj′=(Rj,Rj′,s,t), then the current node N expands into nodes N′ and N″, and the two nodes have all the constraints of N. Adding a new constraint con1=(Rj,s,t) to the child node N′ means that Rj cannot occupy node s at time t. Similarly, adding a new constraint to the child node N″; giving the conflict constraint information to Rj and Rj′ to ensure that one of the four-way shuttle vehicles avoids the conflict and replans the path.

[0030] Preferably, in step 5, the node selection strategy and the minimum turning strategy are used to improve and optimize the lower layer algorithm.

[0031] Furthermore, the node selection strategy is to introduce the actual cost value g(n) into the direction matrix D for optimization adjustment to constrain the selection conditions of the extended nodes;

[0032] The actual cost value after optimization adjustment:

[0033] g′(n)=g(n)+|x c -x n |·(D(x n ,y n )+D(x c ,y c ))·K

[0034] In the formula, (x c ,y c ) represents the coordinate position of the parent node of the expanded node, (x n ,y n ) represents the coordinate position of the current expanded node, K is an integer that approaches infinity, indicating the possibility of node selection. If the calculated value before K is not 0, the cost value of g′(n) approaches infinity and the node cannot be selected.

[0035] Furthermore, the minimum turning strategy is to adjust the actual cost estimation and the heuristic evaluation function to reduce the turning frequency when searching for the shortest path;

[0036] When the algorithm expands the node to a turning node, the actual cost g of the expanded node is increased; the actual cost value from the starting node s to the expanded node n is:

[0037] g'(n)=g(n-1)+c(n-1,n)+sgn(|(x n -x n-2 )(y n -y n-2 )|)·P

[0038] Where c(n-1,n) represents the cost from the current node n-1 to the extended node n without considering the turn; sgn is the sign function; P is the turn cost; (x n ,yn ) and (x n-2 ,y n-2 ) represent the position of the expanded node n and the position of the parent node n-2 of the current node respectively; when turning, x n -x n-2 With y n -y n-2 All are not 0;

[0039] Optimizing the heuristic function includes:

[0040] When the expansion node and the target node are on the same track, there is no turn:

[0041] h′(n)=h(n)+0·P;

[0042] If the expansion node and the target node are not on the same track, there is at least one turn:

[0043] h′(n)=h(n)+1·P;

[0044] If the expansion node and the target node are on the same track but there is an obstacle between them, there are at least 2 turns:

[0045] h′(n)=h(n)+2·P.

[0046] A single-target mission multi-four-way shuttle path planning and obstacle avoidance system, including a data acquisition module, a mission model building module, a CBS-ET algorithm solution model module and a path output module;

[0047] The data acquisition module is used to obtain basic information of the storage system and establish a grid map of the four-way shuttle storage system;

[0048] The task model building module is used to build a task model with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task;

[0049] The CBS-ET algorithm solution model module is used to solve the task model, perform path planning, and find a set of conflict-free optimal paths;

[0050] The path output module is used to output the optimal path to the corresponding four-way shuttle; multiple four-way shuttles complete the single-target task according to the optimal path.

[0051] Compared with the prior art, the present invention has the following beneficial technical effects:

[0052] The present invention provides a single-target task multi-four-way shuttle path planning and obstacle avoidance method. For a single-target task involving multiple four-way shuttles with only one starting point and one end point, the upper-level algorithm is used to globally plan the path and detect conflicts. The time is divided into fixed time steps through a discrete time model, which enables the algorithm to accurately calculate the position and state of the four-way shuttle within each time step, thereby further improving the overall operating efficiency and stability. The lower-level algorithm introduces an extended node method and a minimum turning strategy, and uses the A* algorithm to solve, which significantly improves the solution efficiency, reduces the task completion time, and improves the system's work efficiency. It has important practical guiding significance for the scheduling of the four-way shuttle storage system.

[0053] The present invention solves the single-target task path planning and obstacle avoidance problems by introducing the CBS-ET algorithm with node selection strategy and minimum steering strategy. The node selection strategy can constrain the driving direction of the four-way shuttle in different tracks and improve the algorithm's solution efficiency. The minimum steering strategy optimizes the steering frequency of the four-way shuttle, shortens the task time of the four-way shuttle, and improves the system's work efficiency. The present invention improves the algorithm's solution efficiency, shortens the task time of the four-way shuttle, and improves the system's work efficiency, which has certain practical guiding significance for the scheduling of the four-way shuttle storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is the flow chart of CBS-ET algorithm.

[0055] Figure 2 This is a comparison chart of the bottom routes of the AGV and the four-way shuttle rack.

[0056] Figure 3 The effect of turning on the path.

[0057] Figure 4 This is a grid map of a four-way shuttle storage system.

[0058] Figure 5 This is the path planning result of the general CBS algorithm (4 four-way shuttles).

[0059] Figure 6 This is the three-dimensional time coordinate diagram of the general CBS algorithm (4 four-way shuttles).

[0060] Figure 7 This is the path planning result of the CBS-ET algorithm (4 four-way shuttles).

[0061] Figure 8 This is the three-dimensional time coordinate diagram of the CBS-ET algorithm (4 four-way shuttles).

[0062] Fig. 9 This is the path planning result of the general CBS algorithm (6 four-way shuttles).

[0063] Fig.10 This is the three-dimensional time coordinate diagram of the general CBS algorithm (6 four-way shuttles).

[0064] Fig.11 This is the path planning result of the CBS-ET algorithm (6 four-way shuttles).

[0065] Fig.12 This is the three-dimensional time coordinate diagram of the CBS-ET algorithm (6 four-way shuttles).

[0066] Fig.13 This is the path planning result of the general CBS algorithm (8 four-way shuttles).

[0067] Fig.14 This is the three-dimensional time coordinate diagram of the general CBS algorithm (8 four-way shuttles).

[0068] Fig.15 This is the path planning result of the CBS-ET algorithm (8 four-way shuttles).

[0069] Fig.16 This is the three-dimensional time coordinate diagram of the CBS-ET algorithm (8 four-way shuttles). DETAILED DESCRIPTION

[0070] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0071] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0072] A single-target multi-four-way shuttle path planning and obstacle avoidance method of the present invention comprises the following steps:

[0073] Step 1: Determine the basic information of the storage system, including the number of main lanes, the number of sub-lanes, the layout of cargo channel and the number of shuttle vehicles, and establish a grid map of the four-way shuttle storage system, such as Figure 4 .

[0074] Step 2: Establish an optimization model with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task.

[0075] Step 3: Design the CBS-ET algorithm solution model to solve the single-objective task path optimization problem of multiple four-way shuttles on the same floor.

[0076] Step 4: Set the corresponding model parameters and run the model to obtain the corresponding operation paths of different shuttles.

[0077] In step 2, the total time required for the four-way shuttle to complete a single-target task is used as the objective function to establish a model. The specific model establishment is as follows:

[0078]

[0079] in:

[0080] Formula (1) indicates that there are n four-way shuttles assigned with single-target tasks, whose numbers belong to J = {1, 2, ..., n}, and Rj (j∈J) represents the jth four-way shuttle. Rj ={s Rj ,e Rj} is a single-objective task, s Rj ∈G and e Rj ∈G are single-objective tasks C Rj The starting and ending positions of Rj are the starting and ending points of the path Rj. The time is discretized into T = {1, 2, 3, ...}. The path w of Rj Rj It is a mapping from the time set T to the map space G, w Rj (t) represents the position of Rj at time t. At any time, the four-way shuttle has only two behaviors: waiting or moving to a neighboring location. cost(Rj) represents the cost of Rj performing the transport task C. Rj The objective function of the model is to minimize the total time required for all four-way shuttles to complete a single-target task.

[0081] In formula (2), is the time when Rj departs from the starting point, is the time when Rj arrives at the destination.

[0082] The constraints are as follows:

[0083] w Rj (t)∈G (3)

[0084] z(w Rj (t))=0,z∈G (4)

[0085]

[0086]

[0087] w Rj (t)≠w Rj′ (t),j′∈J,j≠j′ (7)

[0088] w Rj (t)≠w Rj′ (t+1)∧w Rj(t+1)≠w Rj′ (t) (8)

[0089] Formula (3) indicates that Rj's path is in map G. Formula (4) indicates that Rj's path does not pass through obstacles, and z is an obstacle in map G. Formula (5) indicates that the start and end points of Rj's path correspond to the start and end positions of the task. Formula (6) indicates that in each time unit, Rj can only move to an adjacent position or wait in place. and Represent the positions of Rj at time t and t+1 respectively. Formula (7) indicates that two four-way shuttles cannot be at the same position on the map at the same time. Formula (8) stipulates that the four-way shuttles cannot exchange positions when moving;

[0090] In step 3, the CBS-ET algorithm is designed to solve the model. The core of the algorithm is to adopt a hierarchical search strategy. The upper-level algorithm performs global path optimization and conflict detection. If a path conflict occurs, the extended node and minimum turning strategy are introduced, and the lower-level algorithm is used to perform local path planning. The high-level and low-level algorithms are continuously iterated, and a set of conflict-free optimal paths are gradually found through the cycle of conflict detection, constraint addition and path replanning.

[0091] The specific steps of the CBS-ET algorithm are as follows: Figure 1 shown.

[0092] (1) Parameter initialization: Establish a warehouse environment map; set the number, starting and ending positions of the four-way shuttles; set the motion parameters of the four-way shuttles, etc.

[0093] (2) Global planning of the upper-level algorithm: The upper-level algorithm plans the initial path set of multiple four-way shuttles. This stage only generates the initial solution, and the coordination between the four-way shuttles is not considered, so these initial paths may conflict.

[0094] (3) Conflict detection: When multiple four-way shuttles traverse different edges and reach the edge vertices at the same time, a node conflict will occur. When multiple four-way shuttles traverse the same edge, the four-way shuttles moving along the edge will conflict with the four-way shuttles moving in the opposite direction. If no conflict occurs, go to step (5).

[0095] (4) Add constraints. If there is a conflict that Rj and Rj′ (j, j′∈J) occupy a node s at the same time t, it is recorded as (Rj, Rj′, s, t). <v j ,v j′ >, then it is recorded as (Rj,Rj′,v j ,v j′ ,t). If Rj,Rj′ have a node conflict c jj′=(Rj,Rj′,s,t), then the current node N expands to nodes N′ and N″, and the two nodes have all the constraints of N. Adding a new constraint con1=(Rj,s,t) to the child node N′ means that Rj cannot occupy node s at time t. Similarly, adding a new constraint to the child node N″. Give the conflict constraint information to Rj and Rj′ to ensure that one of the four-way shuttles avoids the conflict and replans the path.

[0096] (5) Local planning of lower-level algorithms

[0097] The lower-level path search algorithm replans the path for the constraint node N′ or N″. The four-way shuttle can only travel along the track and cannot travel diagonally or in a curve, so the Manhattan distance is selected as the heuristic function:

[0098] h(n)=|x n -x e |+|y n -y e | (9)

[0099] Among them, (x n ,y n ) is the coordinate of the current node, (x e ,y e ) are the coordinates of the target node.

[0100] As the lower-level algorithm of the CBS algorithm, the node expansion method of the ordinary A* algorithm is not suitable for the four-way shuttle warehouse system map environment, and the number of turns during the path finding process is not constrained. The specific contents of the improvement of the A* algorithm are as follows:

[0101] 1) Node selection strategy, in the four-way shuttle storage system, the four-way shuttle has certain limitations in its driving direction, and it cannot travel in four directions at any position of the storage shelf. Figure 2 As shown in the figure (the square in the figure represents a single cargo location), the AGV can freely travel in four directions at the bottom of the shelf, while the four-way shuttle can only travel longitudinally when traveling at the bottom of the shelf. This is because the four-way shuttle storage vehicle system adopts a stacked storage cargo storage method. The four-way shuttle can travel in four directions in the aisles of the shelves, but can only travel along the shelf sub-tracks in the cargo channel and cannot cross the cargo channel. Therefore, when expanding the node, the A* algorithm based on the four-way shuttle path optimization needs to consider the location of the current node and determine whether all the surrounding 4 nodes can be selected.

[0102] In the four-way shuttle route map, when the starting point expands the node, only nodes ① and ③ can be selected (nodes ② and ④ are unreachable and cannot be selected). Therefore, a direction matrix is ​​introduced for the actual cost value g(n) to constrain the selection conditions of the expanded node. Figure 2 Taking the four-way shuttle route map as an example, the direction matrix is ​​established as follows:

[0103]

[0104] In the formula, 0 means that when the node is expanded at this location, the current node can be expanded in four directions: upward, downward, left, and right, that is, the four-way shuttle at this location can travel in four directions; and 1 means that the current node is limited to the choice of the two directions of upward or downward when expanding, that is, the four-way shuttle can only move longitudinally at this location. By optimizing and adjusting the direction matrix D, the actual cost value of the A* algorithm is redefined:

[0105] g′(n)=g(n)+|x c -x n |·(D(x n ,y n )+D(x c ,y c ))·K (11)

[0106] In the formula, (x c ,y c ) represents the coordinate position of the parent node of the expanded node, (x n ,y n ) represents the coordinate position of the current expanded node, K is an integer that approaches infinity, indicating the possibility of node selection. If the calculated value before K is not 0, the cost value of g′(n) approaches infinity and the node cannot be selected.

[0107] 2) Minimum Turning Strategy: The general A* algorithm usually aims to search for the shortest moving path without considering the turning cost. In the four-way shuttle storage system, the turning of the four-way shuttle is completed by exchanging the wheel train, which will bring a high time cost. Figure 3 In the grid map, although the lengths of paths ① and ② are the same, path ② has fewer turns, only one, while path ① needs to turn twice. Therefore, from the perspective of time consumption, path ② is more efficient than path ①. In order to optimize the A* algorithm, its actual cost estimation and heuristic evaluation function are adjusted to reduce the frequency of turns when finding the shortest path.

[0108] When the algorithm expands a node to a turning node, the actual cost g of the expanded node is increased. The actual cost value from the starting node s to the expanded node n is:

[0109] g'(n)=g(n-1)+c(n-1,n)+sgn(|(x n -x n-2 )(y n -y n-2 )|)·P (12)

[0110] Where c(n-1,n) represents the cost from the current node n-1 to the extended node n without considering the turn; sgn is the sign function. P is the turn cost; (x n ,y n ) and (x n-2 ,y n-2 ) represent the position of the expanded node n and the position of the parent node n-2 of the current node. n -x n-2 With y n -y n-2 Both are not 0.

[0111] In order to reduce the algorithm's turning frequency during the path search process, the following measures are taken to optimize the heuristic function:

[0112] ① When the expansion node and the target node are on the same track, there is no turn:

[0113] h′(n)=h(n)+0·P (13)

[0114] ② If the expansion node and the target node are not on the same track, there is at least one turn, as shown in route ② in the four-way shuttle route map:

[0115] h′(n)=h(n)+1·P (14)

[0116] ③ If the extended node and the target node are on the same track, but there is an obstacle between them, there are at least 2 turns:

[0117] h′(n)=h(n)+2·P (15)

[0118] (6) Path update: After improving the local planning of the A* algorithm, the re-planned four-way shuttles need to update their own path information and return to step (3) to perform conflict detection in combination with the upper-level algorithm. The above process is called repeatedly until there is no conflict in the system. The algorithm ends and outputs a set of conflict-free multi-four-way shuttle paths. The algorithm flow is as follows: Figure 1 shown.

[0119] In step 4, write Python code, set the corresponding model parameters, and run it to obtain the path planning and obstacle avoidance of the multi-directional shuttle.

[0120] The embodiment of the present invention also provides a single-target task multi-four-way shuttle path planning and obstacle avoidance system, including a data acquisition module, a task model construction module, a CBS-ET algorithm solution model module and a path output module.

[0121] Among them, the data acquisition module is used to obtain the basic information of the warehousing system and establish a grid map of the four-way shuttle warehousing system.

[0122] The task model building module is used to build a task model with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task.

[0123] The CBS-ET algorithm solution model module is used to solve the task model, perform path planning, and find a set of conflict-free optimal paths.

[0124] The path output module is used to output the optimal path to the corresponding four-way shuttle; multiple four-way shuttles complete the single-target task according to the optimal path.

[0125] In the embodiment of the present invention, a single-target task and multiple four-way shuttle path planning and obstacle avoidance method and system are used. For a single-target task involving multiple four-way shuttles with only one starting point and one end point, the upper-level algorithm is used to globally plan the path and detect conflicts. The time is divided into fixed time steps through a discrete time model, which enables the algorithm to accurately calculate the position and state of the four-way shuttle within each time step, thereby further improving the overall operation efficiency and stability. The lower-level algorithm introduces the expansion node method and the minimum turning strategy, and uses the A* algorithm to solve, which significantly improves the solution efficiency, reduces the task completion time, and improves the system's work efficiency. It has important practical guiding significance for the scheduling of the four-way shuttle storage system.

[0126] Example 1

[0127] The model of a warehousing system is as follows: Figure 4 As shown in the figure, it includes 4 horizontal main lanes, 5 longitudinal sub-lanes and 279 longitudinal cargo channels. The entire system has 2114 locations that can be used to store goods. Each main lane is equipped with a lift on the far left, which is marked in dark green in the figure. In the figure, the white area represents the channel where the four-way shuttle can travel freely, the light green part represents the cargo space loaded with goods, and the green area represents the idle cargo space where the four-way shuttle can move quickly. The algorithm was experimentally verified using Pycharm simulation software. The relevant parameters of the experimental environment are shown in Table 1.

[0128] Table 1 Picking parameter settings

[0129]

[0130]

[0131] To verify the effectiveness of the improved algorithm, three groups of examples of different scales are designed. Group 1 contains 4 four-way shuttles (R1, R2, R3, R4), Group 2 contains 6 four-way shuttles (R1~R4, R5, R6), and Group 3 contains 8 four-way shuttles (R1~R6, R7, R8). Using the CBS-ET algorithm and the general CBS algorithm, path planning and obstacle avoidance experiments are performed on each group of four-way shuttles, and the calculation time and task completion time are recorded. The path colors of R1~R8 are set to green, red, brown, blue, black, cyan, magenta and bright yellow.

[0132] (1) Analysis of the first set of examples

[0133] Given the starting and target positions of four four-way shuttles, the starting point coordinates of R1 are (18,0) and the target point coordinates are (33,36); the starting point coordinates of R2 are (34,2) and the target point coordinates are (10,0); the starting point coordinates of R3 are (8,10) and the target point coordinates are (53,27); the starting point coordinates of R4 are (52,12) and the target point coordinates are (27,19).

[0134] Based on the general CBS algorithm, the path planning and obstacle avoidance experiments were carried out on the group of four four-way shuttles. The results are as follows: Figure 5 , Figure 6 As shown. Combined with the algorithm output results, it can be concluded that: ① R1 and R3 have a node conflict at the coordinate (20,10). By adding the constraint at this node to the two four-way shuttles, R3's path is replanned, and it chooses to drive upward from the 18th lane in advance, forming a path of (18,10)→(18,12)→(20,12), avoiding the node conflict with R1 at (20,10). ② After the replanned path, R3 has a lateral conflict with R1 at the coordinate (20,12). By adding the lateral edge constraint at this location to the two four-way shuttles, R1 chooses to turn left at the coordinate (21,13) and return to the 20th lane, avoiding a secondary conflict with R3.

[0135] Based on the CBS-ET algorithm, the path planning and obstacle avoidance experiments were carried out on the four four-way shuttles. The results are as follows: Figure 7 , Figure 8 As shown in the figure. Combined with the algorithm output results, it can be seen that the path planning and conflict situation of the four-way shuttle have changed after the algorithm is improved: ① R1 and R3 have a longitudinal conflict at the coordinate (20,9). By adding the longitudinal edge constraint at this node to the two four-way shuttles, R3 replanned the path and chose to detour in the first lane, avoiding the conflict with R1 in the 20th lane.

[0136] ②R1 and R3 no longer have a secondary conflict.

[0137] Based on the general CBS algorithm and the improved algorithm in this paper, the experimental results of the first group of 4 four-way shuttles are compared in Table 2. It can be seen that in the case of 4 four-way shuttles, compared with the general CBS algorithm, the number of turns of the four-way shuttles under the CBS-ET algorithm is effectively reduced, and the total time to complete the task is reduced by 8.7%. R2 does not reflect the advantages of the improved algorithm due to its simple path, but for most four-way shuttles, the improved algorithm has a good effect in reducing the number of turns and shortening the task time.

[0138] Table 2 Comparison of route optimization for the first group of four four-way shuttles

[0139]

[0140] (2) Analysis of the second set of examples

[0141] Given the starting and target positions of 6 four-way shuttles. The starting and target positions of R1 to R4 are the same as those of the first group of experiments. On this basis, two more four-way shuttles are added, of which the starting point coordinates of R5 are (58,20) and the target point coordinates are (52,12); the starting point coordinates of R6 are (27,1) and the target point coordinates are (15,20).

[0142] Based on the general CBS algorithm, the path planning and obstacle avoidance experiments were carried out on the group of 6 four-way shuttles. The results are as follows: Fig. 9 , Fig.10 As shown. Combined with the algorithm output results, it can be concluded that: in addition to the conflict between R1 and R3 in the first set of examples, R4 and R5 have a longitudinal conflict at the coordinate (40,17). After adding the longitudinal edge constraint at this node to the two four-way shuttles, R4 replanned the path and chose to detour from the 20th lane, avoiding the conflict with R5 in the 40th lane.

[0143] Based on the CBS-ET algorithm, the group of 6 four-way shuttles were subjected to path planning and obstacle avoidance experiments. The results are as follows: Fig.11 , Fig.12 As shown. Combined with the algorithm output results, it can be concluded that: ① In addition to the node conflict between R1 and R3 in the first set of examples, R1 and R6 have a lateral conflict at the coordinate (23,0). By adding the lateral edge constraint at this node to the two four-way shuttles, R1 gives up going up the 40th lane and turns to the 20th lane in advance, avoiding the conflict with R6 in the 0th lane. ② R4 and R5 have a longitudinal conflict at the coordinate (40,16). By adding the longitudinal edge constraint at this node to the two four-way shuttles, R4 chooses to detour from the 20th lane to reach the target point (27,19).

[0144] like Fig. 9 , Fig.11 As shown in the figure, similar to the first set of examples, the four-way shuttle based on the general CBS algorithm can travel in four directions on the sub-track of the cargo channel, which does not meet the actual working conditions of the four-way shuttle / RS. The four-way shuttle based on the CBS-ET algorithm can travel in four directions on the mother track, but can only travel longitudinally on the sub-track of the cargo channel. After increasing the number of four-way shuttles, the conflict scenario becomes complicated, but the improved algorithm still meets the actual requirement that the four-way shuttle cannot cross the cargo channel.

[0145] Based on the general CBS algorithm and the improved algorithm in this paper, the experimental results of the second group of 6 four-way shuttles' path planning are compared in Table 4-5. It can be seen that in the case of 6 four-way shuttles, compared with the general CBS algorithm, the number of turns of the four-way shuttles under the CBS-ET algorithm is effectively reduced, and the total time to complete the task is reduced by 10.4%. For most four-way shuttles, the improved algorithm can effectively reduce the number of turns, and the improved algorithm has a greater optimization effect on the total completion time of the task.

[0146] Table 3 Comparison of route optimization for the second group of 6 four-way shuttles

[0147]

[0148] (3) Analysis of the third set of examples

[0149] Given the starting and target positions of 8 four-way shuttles. The starting and target positions of R1 to R6 are the same as those of the previous two experiments. On this basis, two more four-way shuttles are added, of which R7 has a starting point coordinate of (46,20) and a target point coordinate of (20,0); R8 has a starting point coordinate of (61,39) and a target point coordinate of (30,10).

[0150] Based on the general CBS algorithm, the path planning and obstacle avoidance experiments were carried out on the group of 8 four-way shuttles. The results are as follows: Fig.13 , Fig.14 Combined with the algorithm output results, it can be concluded that: since the four-way shuttle has no limitation on the driving direction under the traditional algorithm, after adding two four-way shuttles, except for the node conflict and opposite conflict between R1 and R3 in the first and second groups of examples, and the opposite conflict between R4 and R5, no new conflict occurs in the system.

[0151] Based on the CBS-ET algorithm, the group of 8 four-way shuttles were subjected to path planning and obstacle avoidance experiments. The results are as follows: Fig.15 , Fig.16As shown. Combined with the algorithm output results, it can be concluded that, in addition to the node conflict between R1 and R3 and the conflict between R4 and R5 in the first and second groups of examples: ① R3 and R4 have a lateral conflict at the coordinate (31,9). After adding the lateral edge constraint at the node to the two four-way shuttles, R3 chooses to detour from the first lane and avoids traveling in the opposite direction with R4 in the ninth lane. ② R4 and R7 have a longitudinal conflict at the coordinate (20,10). After adding the longitudinal edge constraint at the node to the two four-way shuttles, R7 chooses to turn down at the coordinate (40,18) and travel in the 40th lane, avoiding traveling in the opposite direction with R4 in the 20th lane. ③ Due to the re-planning of the routes of multiple four-way shuttles, the conflict between R1 and R6 at the coordinate (23,0) no longer occurs.

[0152] like Fig.13 , Fig.15 As shown in the figure, similar to the previous two examples, the four-way shuttle based on the CBS-ET algorithm can travel in four directions on the mother track, but can only travel longitudinally in the sub-track of the cargo channel. After increasing the number of four-way shuttles, the improved algorithm still meets the actual requirement that the four-way shuttle cannot cross the cargo channel.

[0153] The comparison of the experimental results of the path planning of the third group of 8 four-way shuttles based on the general CBS algorithm and the improved algorithm in this paper is shown in Table 4. It can be seen that in the case of 8 four-way shuttles, compared with the general CBS algorithm, the number of turns of the four-way shuttles under the CBS-ET algorithm is effectively reduced, and the total time to complete the task is reduced by 8.8%. For most four-way shuttles, the improved algorithm has shown positive results in reducing the turning frequency and reducing the task execution time, and the optimization effect of the overall task completion time is outstanding.

[0154] Table 4 Path optimization comparison of the third group of 8 four-way shuttle vehicles

[0155]

[0156]

[0157] By analyzing three groups of examples with different numbers and sizes, the performance of the general CBS algorithm and the CBS-ET algorithm in terms of task completion time is compared. It is confirmed that the CBS-ET algorithm can not only meet the different requirements of the four-way shuttle when actually driving on the mother / child track, but also has certain advantages in reducing the number of turns of the four-way shuttle and shortening the total task completion time. As the number of four-way shuttles increases, the optimization rate of the total task time of the improved algorithm is maintained. However, in addition to task time, calculation time is also a key indicator to measure the effect of the path planning algorithm. In order to control the experimental environment with the same conditions, the experiment is still carried out on the above three groups to ensure the universality and accuracy of the evaluation results. The average calculation time required for the general CBS algorithm and the CBS-ET algorithm to solve the same task under different groups is recorded, and the results are shown in Table 5.

[0158] In the path optimization problem, the number of extended nodes is closely related to the algorithm's computational time. The reduction in the number of extended nodes often means that the algorithm only needs to perform fewer computational steps, which will reduce the overall computational time. As shown in Table 4, the improved algorithm has a significant effect in reducing the number of extended nodes, and the computational efficiency also shows a huge advantage. As the number of four-way shuttles increases, the advantage gradually increases. The computational time optimization rates of the three groups are 42.4%, 44.4% and 47.5% respectively. After the improved algorithm adds an extended node selection strategy, the algorithm cannot reach the nodes on the adjacent cargo sub-tracks when expanding the nodes on the cargo sub-tracks. This greatly reduces the number of nodes selected when the algorithm is expanded, improves the algorithm's search efficiency, and also meets the actual working conditions in the four-way shuttle storage system where the four-way shuttle cannot cross the cargo channel.

[0159] Table 5 Comparison of calculation results between general CBS algorithm and CBS-ET algorithm

[0160]

[0161] The present invention proposes a single-target task multiple four-way shuttle path planning and obstacle avoidance method and system, introduces the CBS-ET algorithm with node selection strategy and minimum turning strategy to solve the single-target task path planning and obstacle avoidance problems, optimizes the turning frequency of the four-way shuttle, and thus effectively reduces the conflicts between the four-way shuttles, shortens the task time of the four-way shuttle, and improves the stability and operation efficiency of the warehousing system.

Claims

1. A single-target multi-four-way shuttle path planning and obstacle avoidance method, It is characterized in that The following steps are involved: Obtain basic information of the storage system and establish a grid map of the four-way shuttle storage system; A task model is established with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task; A CBS-ET algorithm solution model is established to solve the task model and perform path planning. The CBS-ET algorithm solution model adopts a hierarchical search strategy. The upper-level algorithm plans multiple initial path sets of four-way shuttles in the grid map of the four-way shuttle warehouse system and performs conflict detection on the initial path sets. When a path conflict occurs, the lower-level algorithm is used to perform local planning on the path and change the optimized path. The upper-level algorithm and the lower-level algorithm are continuously iterated, and a set of conflict-free optimal paths are gradually found through a cycle of conflict detection and path replanning. Multiple four-way shuttles avoid obstacles according to the optimal path and complete single-target tasks.

2. According to claim 1, a single-target task multi-four-way shuttle path planning and obstacle avoidance method, It is characterized in that The basic information of the warehousing system includes the number of main lanes, the number of sub-lanes, the layout of cargo channel layouts and the number of shuttle vehicles.

3. According to the single-target task multi-four-way shuttle path planning and obstacle avoidance method of claim 1, It is characterized in that The task model is: in: is the time when Rj departs from the starting point, is the time when Rj arrives at the destination; n is the number of four-way shuttles assigned with single-target tasks, and its number belongs to J={1,2,...,n}, Rj(j∈J) represents the jth four-way shuttle; C Rj ={s Rj ,e Rj } is a single-objective task, s Rj ∈G and e Rj ∈G are single-objective tasks C Rj The starting and ending positions of Rj are the starting and ending points of the path Rj; cost(Rj) represents the cost of Rj performing the transport task C. Rj The objective function of the task model is to minimize the total time required for all four-way shuttles to complete a single-target task.

4. According to claim 1, a single-target task multi-four-way shuttle path planning and obstacle avoidance method, It is characterized in that After establishing the task model, constrain the task model; The constraints include: w Rj (t)∈G; z(w Rj (t))=0,z∈G; w Rj (t)≠w Rj′ (t),j′∈J,j≠j′; w Rj (t)≠w Rj′ (t+1)∧w Rj (t+1)≠w Rj′ (t); Among them, w Rj (t)∈G represents the path of Rj in the grid map of the four-way shuttle warehouse system; z(w Rj (t)) = 0, z∈G means that the path of Rj does not pass through the obstacle point, and z is the obstacle point in the grid map of the four-way shuttle warehouse system; Indicates the starting and ending points of the Rj path corresponding to the starting and ending positions of the task; Indicates that in each time unit Rj can only move to the adjacent position or wait in place, and Respectively represent the position of Rj at time t and t+1; w Rj (t)≠w Rj′ (t), j′∈J, j≠j′ means that two four-way shuttles cannot be at the same position on the map at the same time; w Rj (t)≠w Rj′ (t+1)∧w Rj (t+1)≠w Rj′ (t) indicates that the four-way shuttle cannot interchange positions while moving.

5. According to claim 1, a single-target task multi-four-way shuttle path planning and obstacle avoidance method, It is characterized in that Establish the CBS-ET algorithm solution model, solve the task model, and perform path planning; The specific steps include: Step 1: Establish a storage environment map, and initialize the number, starting and ending positions, and motion parameters of the four-way shuttles; Step 2, using the upper layer algorithm to plan a set of initial paths for multiple four-way shuttles in the four-way shuttle warehouse system grid map; Step 3, perform conflict detection on the initial path set in step 2; when a path conflict occurs, execute step 4, and if there is no path conflict, directly output the conflict-free optimal path; Step 4: Add constraints to the multiple four-way shuttles to ensure that one of the four-way shuttles avoids conflicts and replans the path; Step 5: Use the lower-level algorithm to perform local planning on the path and change the optimized path; Step 6: The optimized path returns to step 3 to re-check conflicts, and the above steps are repeated until there are no more path conflicts, and the conflict-free optimal path is output.

6. A single-target multi-four-way shuttle path planning and obstacle avoidance method according to claim 5, It is characterized in that In step 4, if there is R j ,R j ′(j,j′∈J) occupy a node s at the same time t, then it is recorded as (Rj,Rj′,s,t); if there is a conflict where Rj and Rj′ travel in opposite directions through an edge at the same time <v j ,v j′ >, then it is recorded as (Rj,Rj′,v j ,v j′ ,t); if Rj,Rj′ has a node conflict c jj′ =(Rj,Rj′,s,t), then the current node N expands to nodes N′ and N″, and the two nodes have all the constraints of N. Add a new constraint con to the child node N′ 1 =(Rj,s,t) means that Rj cannot occupy node s at time t. Similarly, a new constraint is added to the child node N″; the conflict constraint information is given to Rj and Rj′ to ensure that one of the four-way shuttle vehicles avoids the conflict and replans the path.

7. According to claim 1, a single-target task multi-four-way shuttle path planning and obstacle avoidance method, It is characterized in that In step 5, the node selection strategy and minimum turning strategy are used to improve and optimize the lower-level algorithm.

8. A single-target multi-four-way shuttle path planning and obstacle avoidance method according to claim 7, It is characterized in that The node selection strategy is to introduce the actual cost value g(n) into the direction matrix D for optimization adjustment to constrain the selection conditions of the extended nodes; The actual cost value after optimization adjustment: g′(n)=g(n)+|x c -x n |·(D(x n ,y n )+D(x c ,y c ))·K In the formula, (x c ,y c ) represents the coordinate position of the parent node of the expanded node, (x n ,y n ) represents the coordinate position of the current expanded node, K is an integer that approaches infinity, indicating the possibility of node selection. If the calculated value before K is not 0, the cost value of g′(n) approaches infinity and the node cannot be selected.

9. A single-target multi-four-way shuttle path planning and obstacle avoidance method according to claim 7, It is characterized in that The minimum turning strategy is to adjust the actual cost estimate and the heuristic evaluation function to reduce the turning frequency when searching for the shortest path; When the algorithm expands the node to a turning node, the actual cost g of the expanded node is increased; the actual cost value from the starting node s to the expanded node n is: g′(n)=g(n-1)+c(n-1,n)+sgn(|(x n -x n-2 )(y n -y n-2 )|)·P Where c(n-1,n) represents the cost from the current node n-1 to the extended node n without considering the turn; sgn is the sign function; P is the turn cost; (x n ,y n ) and (x n-2 ,y n-2 ) represent the position of the expanded node n and the position of the parent node n-2 of the current node respectively; when turning, x n -x n-2 With y n -y n-2 All are not 0; Optimizing the heuristic function includes: When the expansion node and the target node are on the same track, there is no turn: h′(n)=h(n)+0·P; If the expansion node and the target node are not on the same track, there is at least one turn: h′(n)=h(n)+1·P; If the expansion node and the target node are on the same track but there is an obstacle between them, there are at least 2 turns: h′(n)=h(n)+2·P.

10. A single-target multi-four-way shuttle path planning and obstacle avoidance system, It is characterized in that It includes data acquisition module, task model building module, CBS-ET algorithm solution model module and path output module; The data acquisition module is used to obtain basic information of the storage system and establish a grid map of the four-way shuttle storage system; The task model building module is used to build a task model with the goal of minimizing the total time required for all four-way shuttles to complete a single-target task; The CBS-ET algorithm solution model module is used to solve the task model, perform path planning, and find a set of conflict-free optimal paths; The path output module is used to output the optimal path to the corresponding four-way shuttle; multiple four-way shuttles avoid obstacles according to the optimal path to complete the single-target task.

Images