Logistics sorting system path planning method based on multiple AGVs

Abstract

The invention discloses a logistics sorting system path planning method based on multiple AGVs, and the method comprises the steps: 1, carrying out the modeling of a static environment of a logistics sorting system based on the AGV through a grid method, constructing a directed weighted network diagram on the basis of a grid model, and carrying out the initialization of the weight of a directed arc; 2, dynamically updating the weight of the directed arc; 3, a starting point and a terminal point are distributed to the AGV, a global optimal path of the AGV is generated through an improved labelcorrecting algorithm, and the optimal path of the AGV is dynamically updated; and 4, each automatic guided vehicle AGV moves to the target point according to the distributed path, and dynamically avoids obstacles in the moving process. According to the method, the static grid environment is associated with the directed network weighting graph, and a new construction method of the directed network weighting graph is provided, so that the AGV does not advance towards the arc in the annular deadlock, and the efficiency of searching the optimal path of the AGV is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle path planning, and in particular relates to a method for constructing a directed weighted network graph in a static environment and an AGV (Automated Guided Vehicle) dynamic path planning method based on an improved label_correcting algorithm. Background Art

[0002] The intelligent logistics sorting system based on AGV plays an important role in improving the structure of logistics transportation system, reducing logistics transportation costs, and improving system operation efficiency. It has great application value in logistics sorting system. The collaborative path planning, obstacle avoidance and deadlock problem solving of multiple AGVs have always been the hot spots and difficulties in this research field. A large number of algorithms have been used for AGV path planning in intelligent logistics sorting system. Commonly used algorithms include A * Algorithms, ant colony algorithms, etc. However, the existing calculation methods for AGV driving paths often ignore the calculation of AGV when it turns or has a circular deadlock, so that the path is often not the most optimized path. For example, the traditional A* algorithm mainly focuses on finding the shortest straight-line path. However, in practical applications, AGVs need to accelerate and decelerate when turning, which greatly increases the time cost. If the algorithm does not fully consider the time cost of turning, it may result in the planned path having the shortest straight-line distance but a long overall time consumption, so it is not the optimal path. At the same time, due to the simultaneous operation of multiple AGVs, a circular deadlock may occur, that is, multiple AGVs wait for each other to release resources and cannot move forward. Existing path planning algorithms often do not have a special mechanism to deal with this circular deadlock, resulting in low system efficiency or even complete stagnation. Summary of the invention

[0003] Purpose of the invention: The purpose of the present invention is to provide a path planning method for a logistics sorting system based on multiple AGVs. This method proposes a new method for constructing a directed weighted network graph based on a static environment, and proposes a new method for calculating the weight of directed arcs. The global optimal path of the AGV is calculated using an improved label_correcting algorithm based on the directed weighted network graph. Since the weight calculation takes into account the influence of existing route information and deadlock, the AGV will not move forward to the arc in the circular deadlock, giving the AGV in the deadlock state sufficient time to adjust the path, unlock the circular deadlock, and improve the stability and operation efficiency of the system.

[0004] Technical solution: A path planning method for a logistics sorting system based on multiple AGVs of the present invention comprises the following steps:

[0005] Step 1: Model the static environment of the logistics sorting system based on the automated guided vehicle (AGV) using the grid method, construct a directed weighted network graph based on the grid model, and initialize the weights of the directed arcs;

[0006] Step 2: Dynamically update the weights of directed arcs;

[0007] Step 3: Assign the starting and ending points to the automatic guided transport vehicle AGV, generate the global optimal path of the AGV by improving the label correction algorithm label_correcting, and dynamically update the optimal path of the AGV;

[0008] Step 4: Each automated guided vehicle AGV moves toward the target point according to the assigned path, dynamically avoiding obstacles during the movement.

[0009] Furthermore, step 1 specifically includes the following steps:

[0010] Step 1.1, set the grid points in the same row and column as the sorting entrance as nodes, and record the position coordinates of the nodes;

[0011] Step 1.2, connect each node with a directed arc arrow according to the actual driving route in the static grid network, and record the grid point information contained in each arc;

[0012] Step 1.3: Label the nodes and arcs. The information contained in the nodes includes the node number, node coordinates, arcs entering the node, and arcs leaving the node. The information contained in the arcs includes: arc number, arc starting point number, arc end point number, arc weight, and grid point coordinates contained in the arc.

[0013] Step 1.4: Establish a directed weighted network diagram based on the above information.

[0014] Furthermore, in step 2, the weight calculation formula for dynamically updating the directed arc is:

[0015]

[0016] Where (i, j) represents a directed arc starting from node i and ending at node j; represents the weight of arc (i, j) in time period t; l (i,j) represents the length of arc (i, j); represents the average speed of all AGVs that have passed through arc (i, j) in time period t; n represents the number of turns required to move from node i to node j along arc (i, j); α represents The influence factor of n, β represents the influence factor of n, α+β=1; γ is the influence factor of circular deadlock, γ=0 or ∞, if the arc contains an AGV in a circular deadlock state, then γ=∞, otherwise γ=0; the weight of the arc is not fixed, but changes dynamically according to the AGV passing information, and is updated every t time period;

[0017] Initialize the arc weights. There is no AGV running in the initial network. The average speed of each arc is the same, and the maximum driving speed of the AGV is taken, γ = 0.

[0018] Further, the value of γ is determined, that is, whether the arc contains an AGV in a deadlock state, and the steps include:

[0019] Step 2.1: The current node of AGV is s curr , whose next node is s next , the temporary array loop is empty, and s curr Add to loop;

[0020] Step 2.2, determine node s next Is there an AGV on the node s? next If there is no AGV, the current AGV is not in a deadlock state, and γ = 0; if there is, then s next Add to the loop and go to step 2.3;

[0021] Step 2.3, determine s next Is it in the loop? If it is in the loop, it means a circular deadlock has been formed, and γ = ∞; if not, record s next The next node of the previous AGV is s next2 , will s next Add to the loop and let s curr =s next ,s next =s next2 , go to step 2.2.

[0022] Furthermore, step 3 is specifically as follows: use the improved label_correcting algorithm to calculate and store the preceding nodes from each target point to each grid point in the grid network; the AGV searches for the optimal path to the target point in reverse from the current grid point based on the target point information; dynamically updates the optimal path every t time period; moves to the sorting exit along the optimal path; and uses the ring recovery algorithm to solve the AGV that has been detected to be in a ring deadlock.

[0023] The label_correcting algorithm is a commonly used algorithm for solving the shortest path problem. It labels all nodes in the network through an iterative method. Each iteration does not necessarily convert any vertex label from a temporary label to a permanent label. It only corrects the temporary label once, and all vertex labels are still temporary labels. At the end of the iteration, all vertex labels are converted to permanent labels at the same time. When the label_correcting algorithm is used for path planning, it is necessary to determine the weight of each arc in the running network. The present invention improves the weight of directed arcs in the AGV running network in the algorithm, introduces parameters such as time, number of turns, and circular deadlock influencing factor to minimize the time for AGV to deliver packages. The specific calculation formula and steps are shown in step 1.4.

[0024] In an exemplary embodiment according to the present invention, in step 3, the global optimal path of the AGV is calculated according to the weight of the arc using an improved label_correcting algorithm. Since there are many AGVs in the logistics sorting system, but relatively few sorting entrances, the present invention uses an improved label_correcting algorithm to calculate and store the preceding nodes from each target point to each grid point in the grid network. The AGV searches for the optimal path to the target point in reverse from the current grid point based on the target point information. For the AGV in a circular deadlock, it is determined in turn whether there is an arc at the node where the AGV in the element is located with no AGV at the endpoint or the AGV is not in a circular congestion state. If so, the improved label_correcting algorithm is used to replan its path. The global optimal path of the AGV is updated once every fixed update step size.

[0025] Every fixed update step, collect statistics on the information of AGV passing through the directed arc;

[0026] Calculate the average speed of each arc during the update period based on the information, and record the directed arcs in the ring deadlock state;

[0027] Update the arc weights based on the above information;

[0028] The improved label_correcting algorithm is used to calculate the global optimal path of AGV.

[0029] The weight of each arc is updated according to the information of past AGV driving, so as to dynamically update the optimal path of each target point.

[0030] If an AGV detects another AGV in front of it during its movement, it will stop and wait for the AGV in front to leave before continuing to drive. When an AGV crosses and collides at an intersection, it will choose one of the AGVs to pass according to probability.

[0031] The present invention also discloses a computer device, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method of the present invention.

[0032] The present invention also discloses a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of the method of the present invention are implemented.

[0033] The present invention also discloses a computer program product, comprising a computer program / instruction, which implements the steps of the method of the present invention when executed by a processor.

[0034] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0035] 1. The present invention links a static grid environment with a directed network weighted graph and proposes a new method for constructing a directed network weighted graph;

[0036] 2. The weight of the arc is determined based on the past AGV driving information and the number of turns of the arc and whether it is in a circular deadlock. The influence of the existing route information and the deadlock is fully considered, so that the AGV will not move towards the arc in the circular deadlock, giving the AGV in the deadlock state sufficient time to adjust the path and unlock the circular deadlock;

[0037] 3. The information of the arc is not fixed, but changes dynamically with the driving information of the AGV. The update step size can be set arbitrarily to obtain the dynamic information of the AGV on the arc, which can reduce congestion.

[0038] 4. The label_correcting algorithm can be used to calculate the optimal path from the target point to each grid point. The AGV only needs to call the optimal path based on the current position and the target point position, thereby improving the efficiency of the AGV in finding the optimal path. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Overall flow chart;

[0040] Figure 2 Static environment grid map;

[0041] Figure 3 Directed weighted network graph;

[0042] Figure 4 Update the directed arc weight flow graph;

[0043] Figure 5 Improve label_correcting algorithm flowchart;

[0044] Figure 6 Ring detection flow chart;

[0045] Figure 7 Ring recovery flow chart. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the embodiments of the present invention more clear, various implementation modes of the present invention will be further described in detail below with reference to the accompanying drawings.

[0047] Figure 1 The flow chart of an embodiment of the present invention is shown, and each implementation step is described in detail below.

[0048] Step 101: Modeling the static environment of the AGV-based logistics sorting system using a grid method, constructing a directed weighted network graph based on the grid model, and initializing the weights of the directed arcs;

[0049] In this step, the corresponding two-dimensional coordinates are matched to each grid according to the known static environment information, and the feasible direction of the grid point is determined, the obstacles, sorting entrances and sorting exits are marked, and the speed of AGV movement is set. Figure 2 In this simplified grid model of the logistics sorting system, 201 represents fixed obstacles, such as fences and walls; 202 represents the sorting exit, where the AGV flips the parcel it carries and puts it into the sorting exit; 203 represents the sorting entrance, which is the starting point of the AGV. The target sorting exit is assigned to the AGV here, and the sorting exit is not changed during the movement; 204 represents the movable direction of the current row or column, which can be set arbitrarily according to the actual system requirements.

[0050] After the static environment is modeled using the grid method, a directed weighted network graph is constructed based on the established grid model. Figure 3 That is, Figure 2 A directed weighted network diagram constructed on the basis of , 301 represents a node, and 302 represents a sorting outlet.

[0051] Set the sorting entrance and the grid points in the same row and column as the sorting entrance as nodes, and record the position coordinates of the nodes;

[0052] Connect each node with directional arrows (directed arcs) according to the actual driving route in the static grid network, and record the grid point information contained in each arc;

[0053] The nodes and arcs are labeled. The information contained in the nodes includes the node number, the node coordinates, the arcs entering the node, and the arcs leaving the node. The information contained in the arcs includes: the arc number, the arc starting point number, the arc end point number, the arc weight, and the grid point coordinates contained in the arc.

[0054] Based on the above information, establish a complete directed weighted network graph.

[0055] Step 102: Assign a start point and a destination point to the AGV;

[0056] The AGV is randomly assigned a sorting entrance and a target sorting exit. During the movement of the AGV, no matter how the path changes, the sorting exit is always fixed and unchanged.

[0057] Step 103: Update the weight of the directed arc;

[0058] Figure 4 The flowchart for updating the weight of directed arcs is shown. The weight calculation formula of directed arcs is:

[0059]

[0060] Where (i, j) represents a directed arc starting from node i and ending at node j; represents the weight of arc (i, j) in time period t; l (i,j) represents the length of arc (i, j); It represents the average speed of all AGVs that have passed through arc (i, j) in time period t. When there is no AGV on the arc in time period t, the average speed is the maximum driving speed of the AGV; n represents the number of turns required to move from node i to node j along arc (i, j). α is The influence factor of n, β is the influence factor of n, α+β=1. γ is the influence factor of the ring congestion, γ=0 or ∞, if the arc contains an AGV in a ring deadlock state, then γ=∞, otherwise γ=0; the weight of the arc is not fixed, but changes dynamically according to the AGV passing information, and is updated every t time period.

[0061] Initialize the arc weights. There is no AGV running in the initial network. The average speed of each arc is the same, and the maximum speed of the AGV is taken, and γ = 0.

[0062] According to an exemplary embodiment of the present invention, referring to Figure 6 Determine the value of γ, that is, detect whether the arc contains an AGV in a deadlock state, and the rule is:

[0063] Step 103-1: The current node of the AGV is s curr , whose next node is s next , the temporary array loop is empty, and s curr Add to loop;

[0064] Step 103-2: Determine node s next Is there an AGV on the node s? next If there is no AGV, the current AGV is not in a deadlock state, and γ = 0; if there is, then s nextJoin the loop and go to step 103-3;

[0065] Step 103-3: Determine s next Is it in the loop? If it is in the loop, it means a circular deadlock has been formed, and γ = ∞; if not, record s next The next node of the previous AGV is s next2 , will s next Add to the loop and let s curr =s next ,s next =s next2 , go to step 103-2;

[0066] Step 104: Generate the global optimal path of AGV by improving the label_correcting algorithm and

[0067] The optimal path of AGV is dynamically updated;

[0068] Step 104-1: Initialize, create a linked list LIST, add the starting point s to the linked list LIST, set the starting point s to be labeled 0, i.e., d(s) = 0; its upstream node is 0, i.e., pred(s) = 0; the remaining nodes are labeled ∞, i.e., d(j) = ∞, (j∈N-{s}), where N is the set of all nodes in the network;

[0069] Step 104-2: Remove element i from LIST, detect the downstream node j of the upstream node i, and perform the following operation: If Then And record the upstream node i of node j, that is, pred(j)=i; if node j is not in the LIST, add the node to the linked list LIST and go to step 104-3;

[0070] Step 104-3: Check whether LIST is empty. If it is empty, output the upstream node of each node and go to step 104-4; if it is not empty, go to step 104-2;

[0071] Step 104-4: trace back from the current node node by node to obtain the global optimal path from the starting point to the target point.

[0072] In order to reduce the amount of calculation and improve the efficiency of path finding in this embodiment, the target point of the AGV is sorted out as the starting point s, the upstream node of each node is recorded, and the global optimal path from the current position of the AGV to the target point is traced back. The update is performed every t time period, and the corresponding optimal path is updated every t time period.

[0073] Step 105: Each AGV moves toward the target point according to the assigned path, dynamically avoiding obstacles during the movement.

[0074] If an AGV detects that there are other AGVs in front of it during its movement, it will stop and wait for the AGV in front to leave before continuing to move. When an AGV crosses and collides at an intersection, one of the AGVs will be selected to pass according to probability. For AGVs that have been detected to be in a ring deadlock, the ring recovery algorithm is used to unlock the ring. Figure 7 The loop recovery flow chart is shown, which specifically traverses the elements in the loop one by one, and determines whether there is an arc at the node where the AGV is located in the element with no AGV at the endpoint or the AGV is not in a ring congestion state. If so, the improved label_correcting algorithm is used to replan its path.

Claims

1. A path planning method for a logistics sorting system based on multiple AGVs, characterized in that: The steps include: Step 1: Model the static environment of the logistics sorting system based on the automated guided vehicle (AGV) using the grid method, construct a directed weighted network graph based on the grid model, and initialize the weights of the directed arcs; Step 2: Dynamically update the weights of directed arcs; Step 3: Assign the starting and ending points to the automatic guided transport vehicle AGV, generate the global optimal path of the AGV by improving the label correction algorithm label_correcting, and dynamically update the optimal path of the AGV; Step 4: Each automated guided vehicle AGV moves toward the target point according to the assigned path, dynamically avoiding obstacles during the movement.

2. A path planning method for a logistics sorting system based on multiple AGVs according to claim 1, characterized in that: Step 1 specifically includes the following steps: Step 1.1, set the grid points in the same row and column as the sorting entrance as nodes, and record the position coordinates of the nodes; Step 1.2, connect each node with a directed arc arrow according to the actual driving route in the static grid network, and record the grid point information contained in each arc; Step 1.3: Label the nodes and arcs. The information contained in the nodes includes the node number, node coordinates, arcs entering the node, and arcs leaving the node. The information contained in the arcs includes: arc number, arc starting point number, arc end point number, arc weight, and grid point coordinates contained in the arc. Step 1.4: Establish a directed weighted network diagram based on the above information.

3. The path planning method for a logistics sorting system based on multiple AGVs according to claim 1 is characterized in that: In step 2, the weight calculation formula for dynamically updating directed arcs is: Where (i, j) represents a directed arc starting from node i and ending at node j; represents the weight of arc (i, j) in time period t; l (i,j) represents the length of arc (i, j); represents the average speed of all AGVs that have passed through arc (i, j) in time period t; n represents the number of turns required to move from node i to node j along arc (i, j); α represents The influence factor of n, β represents the influence factor of n, α+β=1; γ is the influence factor of circular deadlock, γ=0 or ∞, if the arc contains an AGV in a circular deadlock state, then γ=∞, otherwise γ=0; the weight of the arc is not fixed, but changes dynamically according to the AGV passing information, and is updated every t time period; Initialize the arc weights. There is no AGV running in the initial network. The average speed of each arc is the same, and the maximum driving speed of the AGV is taken, γ = 0.

4. The path planning method for a logistics sorting system based on multiple AGVs according to claim 3 is characterized in that: Determining the value of γ, that is, detecting whether the arc contains an AGV in a deadlock state, comprises the following steps: Step 2.1: The current node of AGV is s curr , whose next node is s next , the temporary array loop is empty, and s curr Add to loop; Step 2.2, determine node s next Is there an AGV on the node s? next If there is no AGV, the current AGV is not in a deadlock state, and γ = 0; if there is, then s next Add to the loop and go to step 2.3; Step 2.3, determine s next Is it in the loop? If it is in the loop, it means a circular deadlock has been formed, and γ = ∞; if not, record s next The next node of the previous AGV is s next2 , will s next Add to the loop and let s curr =s next ,s next =s next2 , go to step 2.

2.

5. The path planning method for a logistics sorting system based on multiple AGVs according to claim 1 is characterized in that: Step 3 is as follows: the improved label_correcting algorithm is used to calculate and store the predecessor nodes from each target point to each grid point in the grid network. The AGV searches for the optimal path to the target point in reverse from the current grid point based on the target point information. The optimal path is dynamically updated every t time period, and the AGV moves to the sorting exit along the optimal path. For the AGV that has been detected to be in a ring deadlock, the ring recovery algorithm is used to solve the ring.

6. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method of claim 1.

7. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to claim 1 are implemented.

8. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to claim 1 are implemented.

Images