Method and system for planning paths of multiple four-way vehicles in dense warehouse

Abstract

The invention discloses a path planning method and system for multiple four-way vehicles in a dense warehouse, and is suitable for a single-layer multi-vehicle collaborative operation scene of a four-way vehicle dense warehouse, and the method comprises the steps: constructing a rasterized path planning network topological graph according to the actual single-layer goods shelf layout condition of the dense warehouse; according to the network topological graph, the starting point set and the target point set of each four-way vehicle, constructing and solving a multi-four-way vehicle path planning problem on a discrete time step length; performing post-processing on the coarse-grained scheme according to kinematic characteristics and uncertainty existing in the actual operation process; and state estimation is carried out on the running four-way vehicle, and when the access between the theoretical state and the estimated state is too large, the action execution scheme of each four-way vehicle is updated. According to the method, practical constraints such as kinematics characteristics and operation uncertainty of the four-way vehicles are comprehensively considered, and the overall efficiency of the system under the single-layer multi-vehicle collaborative operation scene of the four-way vehicle dense warehouse can be improved to a certain extent.

Description

Technical Field

[0001] The present invention belongs to the technical field of warehousing and logistics, and in particular relates to a method and system for planning paths for multiple four-directional vehicles in a dense warehouse. Background Art

[0002] A four-way shuttle warehouse is a new type of storage system that utilizes four-way shuttles (abbreviated as four-way vehicles) for inbound and outbound operations. As the core equipment in this type of storage system, four-way vehicles are responsible for handling goods in and out of the warehouse. Compared to stacker cranes and shuttles, four-way vehicles can move in all four directions: forward, backward, left, and right, offering significant advantages in flexibility and efficiency.

[0003] However, in actual operations, the operation of four-way vehicles has kinematic characteristics and uncertainties, which can easily lead to actual operations being far different from planned expectations. First, the movement behavior of a four-way vehicle is constrained by its own kinematic characteristics, which is mainly manifested in the vehicle's speed and acceleration limitations, and the kinematic characteristics of the four-way vehicle are different under different loads. Second, there is a certain degree of uncertainty in the movement process of the four-way vehicle, which is specifically manifested in the following two points. First, due to the lack of speed sensors and the influence of factors such as wheel slippage, the actual carrying speed of the four-way vehicle is difficult to estimate. Second, most four-way vehicles use discretely distributed QR codes to identify the current position. When it is in the sensor blind spot between the QR codes, it cannot be located, making it difficult to accurately perceive the operating position.

[0004] Current path planning methods for four-way dense parking systems often rely on simple rules such as first-come, first-served, resulting in low efficiency when multiple vehicles are working together. Existing theoretical research methods typically assume that all four-way vehicles travel at the same uniform speed and can strictly ensure that the actual execution process is consistent with the plan. These problem modeling is overly idealized and significantly simplifies real-world constraints. Therefore, how to improve operational efficiency while considering real-world constraints such as kinematic characteristics and uncertainty is a pressing technical challenge. Summary of the Invention

[0005] In response to the above technical problems, the present invention proposes a path planning method and system for multiple four-way vehicles in a dense warehouse, which comprehensively considers the kinematic characteristics and uncertainty factors existing in the actual operation process of the four-way vehicles; the proposed path planning system can use the path planning method proposed in the present invention to obtain the operation path of each four-way vehicle on the basis of the configurable construction of the virtual operation scene, thereby realizing the scheme preview, virtual operation and performance analysis of the operation process of multiple four-way vehicles on the same level.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for planning paths for multiple four-way vehicles in a densely populated area comprises the following steps:

[0008] Step 1: Based on the layout of single-layer shelves in dense warehouses, a grid-based path planning network topology diagram is constructed;

[0009] Step 2: Based on the rasterized network topology , Gather at the starting point of each four-way vehicle and target point set , construct and solve the multi-way vehicle routing problem on discrete time steps , and obtain the coarse-grained task execution plan for each four-way vehicle ;

[0010] Step 3: Based on the kinematic characteristics and uncertainties in the actual operation process, a coarse-grained task execution plan is developed. Perform post-processing to obtain a refined action execution plan for each four-way vehicle ;

[0011] Step 4: Each four-way vehicle executes the action plan Complete the task and use the four-way vehicle state estimation method to model the motion instability and sensor measurement errors that exist in the actual operation process, and output the optimal estimate of the four-way vehicle position and speed at the current moment; when a small range of error occurs between the theoretical state and the estimated state, adjust it through appropriate acceleration and deceleration; and when the difference is greater than the threshold, go to steps 2 and 3 to update the action execution plan of each four-way vehicle; then go to step 4 to continue executing the task.

[0012] Furthermore, in step 1, the gridded single-layer path planning network topology graph is defined as: .

[0013] Furthermore, the vertex set Includes all unit cargo cells and unit length driving channels in the dense warehouse; vertex information includes the absolute coordinates of the vertex in the grid network and whether it is occupied by cargo or four-way vehicles; edge set A pair of ordered vertices Indicates that Represents the sequence numbers of different nodes in the graph, satisfying ; Edge set Characterize the feasible movement path of the four-way vehicle and include the track constraints that exist in the actual operation of the four-way vehicle.

[0014] Furthermore, the step 2 includes:

[0015] Step 2.1: Path planning problem on discrete time steps Assume that the continuous time domain is discretized into equally spaced time nodes; at each time node , each four-way vehicle Occupy network topology diagram A vertex in , recorded as Between adjacent time nodes, the four-way car moves from the current vertex to the adjacent vertex, or remains stationary. Conflicts should be avoided when moving the four-way car. The goal is to find a collision-free path from the starting position to the target position for each four-way car, while minimizing the task completion time of all four-way cars.

[0016] Step 2.2, a two-level search path planning method based on lazy constraint addition search; The high-level search of the two-level search path planning method maintains a high-level node list, and each high-level node in the list Includes a configuration relationship and a constraint tree ; Among them, the configuration relationship is a network topology diagram A graph with the same topological structure depicts the vertices occupied by each four-way vehicle at the current moment. The constraint tree is a multi-branch tree in which nodes at different depths represent the position constraints of different four-way vehicles. Each node at the same depth represents the vertex position that a specific four-way vehicle may reach at the next moment, provided that the position constraints of its ancestor nodes are met.

[0017] The low-level search of the two-level search path planning method lazily expands the constraint tree in the high-level nodes according to the breadth-first principle; if the configuration relationship of the node is the same as the target configuration relationship, the node is backtracked and a coarse-grained planning solution is output; if the high-level node list is empty and the configuration relationship of the high-level node is different from the target configuration relationship, no solution is output.

[0018] Furthermore, the step 3 includes:

[0019] Step 3.1: Build a temporal planning diagram ,

[0020] Step 3.2, Expand the temporal planning graph: The first type of edge in the temporal planning graph Add safety markings;

[0021] Step 3.3, Temporal Planning Diagram Convert to a simple temporal network ;

[0022] Step 3.4: Solve the optimization problem: minimize the completion time of all tasks In order to optimize the goal, the timing constraints on all edges are taken as the constraints of the problem, and the events in the simple temporal network are Time of occurrence Optimize and determine the optimal time for each action event .

[0023] Furthermore, the step 3.1 includes the following steps:

[0024] Step 3.1.1. Vertex Derived from the non-stop action nodes in the execution plan: For each four-way vehicle Coarse-grained planning path , remove the stop action nodes and get the non-stop path sequence of the four-way vehicle ,in represents the length of the non-stop path sequence; for Each node in , generate the corresponding event vertex ;

[0025] Step 3.1.2: Construct the first type of edge :for Two nodes at adjacent times and , the corresponding event node and satisfy , that is, the edge between the above two event nodes is included in the temporal planning graph The edge set The same below;

[0026] Step 3.1.3: Construct the second type of edge :Order No. Hedi The non-stop path sequence of the four-way vehicle is: and ,like The position occupied by occupies the same position in , the corresponding event node and satisfy .

[0027] Furthermore, the step 3.3 includes the following steps:

[0028] Step 3.3.1. Each four-way vehicle Determine its maximum carrying speed and acceleration range, respectively and ;

[0029] Step 3.3.2: Calculate the first type of edge Timing constraints on :According to the different motion laws, the process is divided into uniform acceleration and deceleration linear motion process, uniform speed linear motion process and mixed motion process. The uncertainty of the four-way vehicle operation is considered in the calculation process.

[0030] Step 3.3.3. Calculate the second type of edge Timing constraints on :For the second type of edge Timing constraints on ;

[0031] Step 3.3.4, Add the start node and end node , and define the timing constraints of the edges from the start node to all zero-indegree nodes as , and the timing constraints of all edges from zero-outdegree nodes to end nodes are .

[0032] Furthermore, in step 4.1, the time information of the four-way vehicle reaching the current cargo location node is obtained through the virtual sensor;

[0033] Step 4.2: Find the corresponding running section at the current moment in the action execution plan of the four-way vehicle, and calculate the theoretical position of the four-way vehicle at the current moment based on the planned path and kinematic model;

[0034] Step 4.3: The measured position obtained by the sensor and the theoretical position derived from the kinematic model are used as inputs of the four-way vehicle state estimation method, and the state estimation of the four-way vehicle is output.

[0035] Furthermore, the step 4.3 includes the following steps:

[0036] Step 4.3.1. Define the state vector From the current moment Four-way vehicle position and speed Two components; define the control input , its physical meaning is the acceleration of the four-way vehicle, and its value is related to the load borne by the four-way vehicle; the state transfer matrix and control matrix Derived from the dynamics formula; define the process noise covariance matrix , used to characterize the potential uncertainty fluctuations in the system's internal motion model; the optimal state estimate at the previous moment is known and the posterior estimated error covariance ;Through the state prediction formula Get the predicted values of the four-way vehicle position and speed at the current moment , by predicting the estimated covariance update formula Get the covariance matrix of the predicted value at the current moment ;

[0037] Step 4.3.2: The four-way vehicle obtains position readings with measurement errors through discrete position sensors. , define the observation matrix as , the observation noise covariance is ; Through the optimal Kalman gain Taking into account the credibility of the predicted value and the observed value, the optimal position and speed estimate of the four-way vehicle at the current moment is obtained. and the posterior estimation error covariance at the current moment .

[0038] The present invention also provides a system for planning paths for multiple four-way vehicles in a densely populated warehouse, which is used to implement the above-mentioned method for planning paths for multiple four-way vehicles in a densely populated warehouse, and includes the following modules:

[0039] The scenario building module completes the configurable construction of a single-layer virtual operation scenario in a four-way densely populated warehouse and provides a virtual operation scenario. Configurable items include the length and width of the cargo unit, the number of cargo locations, the location of the main aisle, the direction of the cargo aisle, and the location of occupied cargo locations. Virtual sensors are installed at each cargo location to simulate the QR code tag-based positioning method in real-world scenarios.

[0040] The coarse-grained path planning module completes the coarse-grained planning of the operation paths of multiple four-way vehicles. Configurable items include the starting and target positions of each four-way vehicle, and supports online re-planning function.

[0041] The fine-grained path planning module post-processes the coarse-grained action execution plan obtained in the coarse-grained path planning module to generate a fine-grained motion plan. Configurable items include the load of each four-way vehicle, the maximum driving speed of each four-way vehicle, the carrying acceleration range of each four-way vehicle, the probability of speed disturbance, the probability of position disturbance and the disturbance amplitude, and supports online re-planning.

[0042] In the virtual operation module, each four-way vehicle operates in a virtual operation scene according to the action execution plan, obtains the measured position information of the four-way vehicle through virtual sensors, obtains the theoretical state information of the four-way vehicle through the action execution plan, updates the status of each four-way vehicle in real time, and renders its position in real time in the virtual operation scene.

[0043] The beneficial effects of the present invention are embodied in:

[0044] The proposed method for path planning for multiple four-way vehicles in a densely packed warehouse improves the efficiency of their coordinated operations while taking into account real-world constraints such as the kinematic characteristics and operational uncertainty of the vehicles during actual operation. This results in a highly accurate, robust, and efficient four-way vehicle motion execution plan. Furthermore, the proposed system supports the configurable construction of virtual work scenarios and provides a solution preview, virtual operation, and solution evaluation environment for the aforementioned path planning method. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flowchart of a method for planning paths for multiple four-way vehicles in a densely populated area according to the present invention;

[0046] Figure 2 This is a flowchart of a method for solving a discrete time step path planning problem according to the present invention;

[0047] Figure 3 This is a flowchart of the post-processing method of the coarse-grained path planning solution of the present invention;

[0048] Figure 4 This is a block diagram of the architecture of the densely populated warehouse multi-vehicle four-way vehicle path planning system of the present invention. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. The technical solutions proposed by the present invention will be described in detail below with reference to the accompanying drawings.

[0050] like Figure 1 As shown, a method for planning paths for multiple four-way vehicles in a densely packed warehouse according to the present invention includes the following steps:

[0051] Step 1: Based on the actual single-layer shelf layout of the dense warehouse, a gridded single-layer path planning network topology graph G is constructed, which is formally expressed as Among them, the vertex set Indicates the location nodes that can be reached by four-way vehicles, which are composed of all unit cargo cells and unit length driving channels in the dense warehouse. The vertex information includes the absolute coordinates of the vertex in the grid network and whether it is occupied by cargo or four-way vehicles. Edge set Defines the connectivity between vertices, consisting of a pair of ordered vertices Indicates that Represents the sequence numbers of different nodes in the graph, satisfying . Edge set It actually represents the feasible moving path of the four-way vehicle and contains the track constraints that exist in the actual operation of the four-way vehicle, that is, the four-way vehicle can only move along the main channel and the cargo channel, and cannot move on the cargo channel or across the cargo channel.

[0052] Step 2: Based on the rasterized network topology , Gather at the starting point of each four-way vehicle and target point set , construct and solve the multi-four-way vehicle path planning problem on discrete time steps. The problem input is formally described as a triple The specific problem modeling and solution are implemented as follows:

[0053] Step 2.1, problem modeling. The path planning problem on discrete time steps assumes that the continuous time domain is discretized into equally spaced time nodes. At each time node , each four-way vehicle Occupy network topology diagram A vertex in , recorded as Between adjacent time nodes, the four-way vehicle can move from the current vertex to the adjacent vertex, or stay at the current vertex. Vertex conflicts and edge conflicts should be avoided when moving the four-way vehicle. Specifically, vertex conflicts refer to the situation where different four-way vehicles occupy the network topology at the same time node. The edge conflict refers to the situation where different four-way vehicles pass through the network topology at the same time. The goal of the problem is to find a collision-free path from the starting position to the target position for each four-way vehicle, while minimizing the task completion time of all four-way vehicles.

[0054] Step 2.2, problem solving. In order to solve the above problem with high quality and high efficiency, the present invention adopts a two-level search path planning method based on lazy constraint addition search. Its high-level search process maintains a high-level node list, and each high-level node in the list Includes a configuration relationship and a constraint tree Among them, the configuration relationship is a network topology diagram The graph with the same topological structure depicts the vertices occupied by each four-way vehicle at the current moment. The constraint tree is a multi-branch tree. Nodes of different depths in the tree represent the position constraints of different four-way vehicles. Nodes of the same depth represent the vertex positions that a specific four-way vehicle may reach at the next moment, provided that the position constraints of the ancestor nodes are satisfied. The low-level search process of this method lazily expands the constraint tree in the high-level nodes according to the breadth-first principle, and finally obtains Coarse-grained task execution scheme for four-way vehicles .in, Describes the Taiwan Four-way Bus starts from the starting point To the end The position of each vertex passed and the arrival time node, n is the total number of four-way vehicles, T is the number of points. The overall process of the algorithm is as follows Figure 2 As shown, the specific solution process is as follows:

[0055] Step 2.2.1, Initialization. Initialize the high-level node list, which contains only one initial high-level node , initial configuration relationship , the initial constraint tree , that is, there is only one root node Used to represent the starting node of the constraint tree.

[0056] Step 2.2.2, high-level search process. High-level search selects a high-level node in the high-level node list according to the depth-first principle. If the configuration relationship of the high-level node is consistent with the target The configuration relationship is the same, that is When , the search is terminated and the node backtracking is performed to obtain the coarse-grained task execution plan of each four-way vehicle. ; If the high-level node list is empty and the configuration relationship of the high-level node is consistent with the target If the configuration relationship is different, the algorithm outputs no solution.

[0057] Step 2.2.3, low-level search process. The low-level search follows the breadth-first principle and searches for the selected high-level nodes. Constraint tree in Perform lazy expansion to generate new configuration relationships that meet the position constraints. Specifically, the low-level search process first selects A leaf node in (parent node), and then randomly select a four-way car without position constraints , and according to the configuration relationship Get a four-way car At the current moment Occupied vertices .

[0058] For four-way vehicles Every possible position at the next moment , generating a pointer to the parent node Finally, according to the configuration relationship and leaf nodes The position constraints of its ancestor nodes generate a new configuration relationship. If this configuration relationship appears for the first time in a high-level node, it is combined with the constraint tree with only one root node and added to the high-level node list. If the configuration relationship of the current node is consistent with the target node, If the configuration relationships are the same, node backtracking is performed, starting from the configuration relationship of the last node and going back to the configuration relationship of the starting node, and finally outputting a coarse-grained planning solution. The low-level search process only expands one leaf node and generates only one configuration relationship that meets the constraint requirements, which is highly efficient.

[0059] Step 3: Execute the task plan obtained after executing step 2 Considering that four-way vehicles with different loads have different motion characteristics and that there are uncertainties such as execution delays and mechanical errors in actual operations, the present invention uses a coarse-grained solution post-processing method based on a temporal planning graph to obtain a more accurate execution solution. Figure 3 The overall flow chart of the post-processing method is as follows:

[0060] Step 3.1: Construct a temporal planning diagram. A temporal planning diagram is a graphical structure used to qualitatively describe the sequence of events. , formally described as Among them, the vertex Describes the four-way vehicle arrival network topology For example, In the coarse-grained solution In, Taiwan's four-way bus Time arrival topology in Position action event. Characterizes the order in which events occur, that is, events Must be earlier than the event Occurs. The plan will be executed Convert to temporal planning diagram The specific implementation scheme is as follows:

[0061] Step 3.1.1: Vertices in the temporal planning graph By implementation plan The non-stop action nodes in the derive. For each four-way vehicle Coarse-grained planning path , remove The stop action node in Non-stop route sequence for four-way vehicles in Taiwan ,in Indicates the length of the non-stop path sequence. Each node in , generate the corresponding event vertex .

[0062] Step 3.1.2: Edges in the Temporal Planning Graph According to the different sources of constraints, they can be divided into two categories. The first type of edge can be directly generated by the non-stop path sequence. Derived to represent Zhongyi four-way vehicle The order of each action event. Two nodes at adjacent times and , the corresponding event node and satisfy , that is, the edge between the above two event nodes is included in the temporal planning graph The edge set The second type of edge describes the order that must be met when different four-way vehicles arrive at the same vertex. Hedi The non-stop path sequence of the four-way vehicle is: and , assuming The position occupied by occupies the same position in , the corresponding event node and satisfy At this point, the temporal planning graph is constructed.

[0063] Step 3.2: Expand the temporal planning graph. Add two safety markers to each first-class edge in the temporal planning graph to describe the minimum safe distance that must be maintained between four-way vehicles, thereby ensuring the safety of multiple four-way vehicles when entering and exiting the same location.

[0064] Step 3.3: Based on the kinematic characteristics and uncertainty of the four-way car, the temporal planning graph is converted into a simple temporal network. A simple temporal network is also a type of graph structure used to describe the sequence of events. , formally described as Among them, the vertex The meaning of is consistent with the vertex meaning in the temporal planning graph. The edges in It can also be divided into two categories, but the difference from the temporal planning graph is that the edges of the simple temporal network are Has a quantitative timing constraint , formally defined as . Its meaning is event Before the event The time interval between the two events ,in Representing an event The time of occurrence, and They represent the upper and lower bounds of the time interval between two events.

[0065] Based on the definition of a simple temporal network and the real-world constraints of a four-way vehicle, this paper associates kinematic factors and uncertainty fluctuations with the quantitative timing constraints on the edges of the simple temporal network. The specific implementation of the simple temporal network is as follows:

[0066] Step 3.3.1, according to factors such as cargo load and driving force, each four-way vehicle Determine its maximum carrying speed and acceleration range, which are defined as and .

[0067] Step 3.3.2: Calculate the first type of edge According to the different motion laws, it can be divided into uniform acceleration and deceleration linear motion process, uniform speed linear motion process and mixed motion process. The specific implementation is as follows:

[0068] The uniform acceleration and deceleration linear motion process is explained in detail using the uniform acceleration linear motion process as an example: Define the connecting edge The length is Assuming a four-way vehicle Able to Within the distance, complete the End speed The uniformly accelerated linear motion process satisfies the acceleration According to the kinematic formula, the four-way vehicle passes Theoretical time satisfy Due to the uncertainty, Timing constraints on ,in is a random variable that obeys an exponential distribution and is used to characterize the extra time required for the four-way car to actually complete the action (the same below).

[0069] Uniform linear motion process: Assume that the four-way vehicle exist within the distance of The speed of the four-way car is to maintain uniform linear motion. According to the kinematics formula, the four-way car passes Theoretical time Due to the uncertainty, Timing constraints on .

[0070] Hybrid motion process: Assuming a four-way vehicle exist Within the distance, first use the known acceleration Completed from arrive The uniformly accelerated linear motion process, and then The speed of the four-way car is to maintain uniform linear motion. According to the kinematics formula, the four-way car passes Theoretical time Due to the uncertainty, Timing constraints on .

[0071] Step 3.3.3. Calculate the second type of edge For the second type of edge Timing constraints on , just ensure Before occurs. Therefore, the definition .

[0072] Step 3.3.4: Add a start node to the front of the simple temporal network , add the end node at the end . And define The timing constraint for edges to all zero-indegree nodes is , all zero-outdegree nodes to The timing constraint of the edge is At this point, the simple temporal network is constructed.

[0073] Step 3.4: Solve the optimization problem to minimize the completion time of all tasks. As the optimization goal of the problem, the timing constraints on all edges are taken as the constraint conditions of the problem, and the events in the simple temporal network are Time of occurrence Optimize and determine the optimal time for each action event Finally, a more refined and robust action execution plan for each four-way vehicle is obtained. .in, Describes the Taiwan Four-way Bus starts from the starting point To the end The positions of each vertex including the safety node and the continuous arrival time, where the superscript Indicates the sequence number of the vertex passed, Used to represent the number of vertices in the action execution plan.

[0074] Step 4: Each four-way vehicle executes the action plan Complete the operation and use a four-way car state estimation method to model the motion instability and sensor measurement errors that exist in the actual operation process, and output the optimal estimate of the four-way car's position and speed at the current moment. The specific implementation method is as follows:

[0075] Step 4.1: Obtain the time information of the four-way vehicle reaching the current cargo location node through the virtual sensor.

[0076] Step 4.2: Find the running section corresponding to the current moment in the action execution plan of the four-way vehicle, and calculate the theoretical position of the four-way vehicle at the current moment based on the planned path and kinematic model.

[0077] Step 4.3: The measured position obtained by the sensor and the theoretical position derived from the kinematic model are used as inputs to the four-way vehicle state estimation method. Preferably, a state estimation method based on Kalman filtering is adopted, and the specific implementation method is as follows:

[0078] Step 4.3.1, prediction phase. Assume that the four-way car maintains uniform acceleration linear motion from the previous position to the current position, meeting the requirements of the linear system. Define the state vector From the current moment Four-way vehicle position and speed Consists of two components.

[0079] ;

[0080] Defining control inputs , its physical meaning is the acceleration of the four-way car, and its value is related to the load borne by the four-way car. According to the dynamic formula of uniformly accelerated linear motion:

[0081] ;

[0082] ;

[0083] The state transition matrix of the system can be deduced and control matrix ,in Indicates the time interval between two locations.

[0084] ;

[0085] ;

[0086] Define the process noise covariance matrix , used to characterize potential uncertain fluctuations such as wheel slip in the system's internal motion model.

[0087] Know the optimal state estimate at the previous moment and the posterior estimated error covariance . Through the state prediction formula Able to obtain the predicted values of the current four-way vehicle position and speed . The covariance update formula is estimated by prediction The covariance matrix of the predicted value at the current moment can be obtained .

[0088] Step 4.3.2, Update phase. The four-way car obtains position readings with measurement errors through discrete position sensors. , define the observation matrix as , and assuming that the observation noise covariance Obey the normal distribution. Through the optimal Kalman gain Taking into account the credibility of the predicted value and the observed value, the optimal position and speed estimate of the four-way vehicle at the current moment is obtained. . And through the formula Get the posterior estimation error covariance at the current moment. The superscript T represents the transpose of the matrix.

[0089] If there's a small discrepancy between the theoretical and estimated states, adjustments can be made through appropriate acceleration and deceleration. If the discrepancy exceeds a certain threshold, the system must proceed to steps 2 and 3 to update the action plans for each of the four-way vehicles. Then, the system must proceed to step 4 to continue executing the task.

[0090] According to another aspect of the present invention, a system for planning paths for multiple four-way vehicles in a densely populated area is provided. The system architecture diagram and the relationship between the modules are shown in FIG. Figure 4 shown.

[0091] The scenario building module allows for the configurable construction of a single-story virtual operation scenario for a four-way, densely packed warehouse, and provides virtual operational scenarios. Configurable items include basic scenario parameters such as the length and width of the storage unit, the number of storage units arranged, the location of the main aisle, the direction of the cargo aisle, the location of occupied storage units, and track constraint parameters. Furthermore, virtual sensors are installed at each storage unit to simulate the QR code tag-based positioning method commonly used in real-world scenarios.

[0092] The coarse-grained path planning module completes coarse-grained planning of the operating paths of multiple four-way vehicles. Configurable items include location parameters such as the starting and target positions of each four-way vehicle, as well as related algorithm parameters, and supports online replanning.

[0093] The fine-grained path planning module post-processes the coarse-grained action execution plans obtained from the coarse-grained path planning module to generate fine-grained motion plans. Configurable parameters include kinematic parameters such as the load of each four-way vehicle, the maximum speed of each four-way vehicle, the acceleration range of each four-way vehicle, the probability of speed disturbance, the probability of position disturbance, and the disturbance amplitude, as well as related algorithm parameters. It also supports online replanning.

[0094] In the virtual operation module, each four-way vehicle operates in a virtual work scenario according to the action execution plan. The module obtains the vehicle's measured position information through virtual sensors and its theoretical state information through the action execution plan. The module updates the vehicle's status in real time and renders its position in the virtual work scenario in real time.

[0095] In summary, the present invention discloses a method and system for planning the paths of multiple four-way vehicles in a densely packed warehouse. Specifically, the path planning method proposed in the present invention comprehensively considers the kinematic characteristics and operational uncertainty factors present during the actual operation of the four-way vehicles. The proposed path planning system utilizes the path planning method proposed in the present invention to determine the operating paths of each four-way vehicle based on the configurable construction of a virtual operation scene, thereby enabling functions such as previewing the plan and virtual operation of multiple four-way vehicles operating on the same floor.

[0096] The contents not described in detail in the specification of the present invention belong to the prior art known to those skilled in the art.

[0097] 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 modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for planning paths for multiple four-way vehicles in a densely populated area, characterized in that: The following steps are involved: Step 1: Based on the layout of single-layer shelves in dense warehouses, a grid-based path planning network topology diagram is constructed; Step 2: Based on the rasterized network topology , Gather at the starting point of each four-way vehicle and target point set , construct and solve the multi-way vehicle routing problem on discrete time steps , and obtain the coarse-grained task execution plan for each four-way vehicle ; Step 3: Based on the kinematic characteristics and uncertainties in the actual operation process, a coarse-grained task execution plan is developed. Perform post-processing to obtain a refined action execution plan for each four-way vehicle ; Step 4: Each four-way vehicle executes the action plan Complete the operation and use the four-way car state estimation method to model the motion instability and sensor measurement errors that exist during the actual operation process, and output the optimal estimate of the four-way car's position and speed at the current moment. When a small range of error occurs between the theoretical state and the estimated state, it adjusts by appropriate acceleration and deceleration. When the difference is greater than the threshold, the process goes to step 2 and step 3 to update the action execution plan of each four-way vehicle; then the process goes to step 4 to continue executing the task.

2. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 1, characterized in that: In step 1, the gridded single-layer path planning network topology graph is defined as: .

3. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 2, characterized in that: Vertex Set Includes all unit cargo cells and unit length driving channels in the dense warehouse; vertex information includes the absolute coordinates of the vertex in the grid network and whether it is occupied by cargo or four-way vehicles; edge set A pair of ordered vertices Indicates that Represents the sequence numbers of different nodes in the graph, satisfying ; Edge set Characterize the feasible movement path of the four-way vehicle and include the track constraints that exist in the actual operation of the four-way vehicle.

4. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 1, characterized in that: The step 2 includes: Step 2.1: Path planning problem on discrete time steps Assume that the continuous time domain is discretized into equally spaced time nodes; at each time node , each four-way vehicle Occupy network topology diagram A vertex in , recorded as Between adjacent time nodes, the four-way car moves from the current vertex to the adjacent vertex, or remains stationary. Conflicts should be avoided when moving the four-way car. The goal is to find a collision-free path from the starting position to the target position for each four-way car, while minimizing the task completion time of all four-way cars. Step 2.2, a two-level search path planning method based on lazy constraint addition search; The high-level search of the two-level search path planning method maintains a high-level node list, and each high-level node in the list Includes a configuration relationship and a constraint tree ; Among them, the configuration relationship is a network topology diagram A graph with the same topological structure depicts the vertices occupied by each four-way vehicle at the current moment. The constraint tree is a multi-branch tree in which nodes at different depths represent the position constraints of different four-way vehicles. Each node at the same depth represents the vertex position that a specific four-way vehicle may reach at the next moment, provided that the position constraints of its ancestor nodes are met. The low-level search of the two-level search path planning method lazily expands the constraint tree in the high-level nodes according to the breadth-first principle; if the configuration relationship of the node is the same as the target configuration relationship, the node is backtracked and a coarse-grained planning solution is output; if the high-level node list is empty and the configuration relationship of the high-level node is different from the target configuration relationship, no solution is output.

5. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 1, characterized in that: The step 3 includes: Step 3.1: Build a temporal planning diagram , Step 3.2, Expand the temporal planning graph: The first type of edge in the temporal planning graph Add safety markings; Step 3.3, Temporal Planning Diagram Convert to a simple temporal network ; Step 3.4: Solve the optimization problem: minimize the completion time of all tasks In order to optimize the goal, the timing constraints on all edges are taken as the constraints of the problem, and the events in the simple temporal network are Time of occurrence Optimize and determine the optimal time for each action event .

6. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 5, characterized in that: The step 3.1 includes the following steps: Step 3.1.

1. Vertex Derived from the non-stop action nodes in the execution plan: For each four-way vehicle Coarse-grained planning path , remove the stop action nodes and get the non-stop path sequence of the four-way vehicle ,in represents the length of the non-stop path sequence; for Each node in , generate the corresponding event vertex ; Step 3.1.2: Construct the first type of edge :for Two nodes at adjacent times and , the corresponding event node and satisfy , that is, the edge between the above two event nodes is included in the temporal planning graph The edge set The same below; Step 3.1.3: Construct the second type of edge :Order No. Hedi The non-stop path sequence of the four-way vehicle is: and ,like The position occupied by occupies the same position in , the corresponding event node and satisfy .

7. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 6, characterized in that: The step 3.3 includes the following steps: Step 3.3.

1. Each four-way vehicle Determine its maximum carrying speed and acceleration range, respectively and ; Step 3.3.2: Calculate the first type of edge Timing constraints on :According to the different motion laws, the process is divided into uniform acceleration and deceleration linear motion process, uniform speed linear motion process and mixed motion process. The uncertainty of the four-way vehicle operation is considered in the calculation process. Step 3.3.

3. Calculate the second type of edge Timing constraints on :For the second type of edge Timing constraints on ; Step 3.3.4, Add the start node and end node , and define the timing constraints of the edges from the start node to all zero-indegree nodes as , and the timing constraints of all edges from zero-outdegree nodes to end nodes are .

8. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 1, characterized in that: The four-way vehicle state estimation method in step 4 comprises the following steps: Step 4.1, obtain the time information of the four-way vehicle reaching the current cargo location node through the virtual sensor; Step 4.2: Find the corresponding running section at the current moment in the action execution plan of the four-way vehicle, and calculate the theoretical position of the four-way vehicle at the current moment based on the planned path and kinematic model; Step 4.3: The measured position obtained by the sensor and the theoretical position derived from the kinematic model are used as inputs of the four-way vehicle state estimation method, and the state estimation of the four-way vehicle is output.

9. The method for planning paths for multiple four-way vehicles in a densely packed warehouse according to claim 8, characterized in that: The step 4.3 includes the following steps: Step 4.3.

1. Define the state vector From the current moment Four-way vehicle position and speed Two components; define the control input , its physical meaning is the acceleration of the four-way vehicle, and its value is related to the load borne by the four-way vehicle; the state transfer matrix and control matrix Derived from the dynamics formula; define the process noise covariance matrix , used to characterize the potential uncertainty fluctuations in the system's internal motion model; the optimal state estimate at the previous moment is known and the posterior estimated error covariance ;Through the state prediction formula Get the predicted values of the four-way vehicle position and speed at the current moment , by predicting the estimated covariance update formula Get the covariance matrix of the predicted value at the current moment ; Step 4.3.2: The four-way vehicle obtains position readings with measurement errors through discrete position sensors. , define the observation matrix as , the observation noise covariance is ; Through the optimal Kalman gain Taking into account the credibility of the predicted value and the observed value, the optimal position and speed estimate of the four-way vehicle at the current moment is obtained. and the posterior estimation error covariance at the current moment .

10. A path planning system for multiple four-way vehicles in a densely populated warehouse, characterized in that: The method for implementing the path planning method for multiple four-way vehicles in a densely populated area according to any one of claims 1 to 9 comprises the following modules: The scenario building module completes the configurable construction of a single-layer virtual operation scenario in a four-way densely populated warehouse and provides a virtual operation scenario. Configurable items include the length and width of the cargo unit, the number of cargo locations, the location of the main aisle, the direction of the cargo aisle, and the location of occupied cargo locations. Virtual sensors are installed at each cargo location to simulate the QR code tag-based positioning method in real-world scenarios. The coarse-grained path planning module completes the coarse-grained planning of the operation paths of multiple four-way vehicles. Configurable items include the starting and target positions of each four-way vehicle, and supports online re-planning function. The fine-grained path planning module post-processes the coarse-grained action execution plan obtained in the coarse-grained path planning module to generate a fine-grained motion plan. Configurable items include the load of each four-way vehicle, the maximum driving speed of each four-way vehicle, the carrying acceleration range of each four-way vehicle, the probability of speed disturbance, the probability of position disturbance and the disturbance amplitude, and supports online re-planning. In the virtual operation module, each four-way vehicle operates in a virtual operation scene according to the action execution plan, obtains the measured position information of the four-way vehicle through virtual sensors, obtains the theoretical state information of the four-way vehicle through the action execution plan, updates the status of each four-way vehicle in real time, and renders its position in real time in the virtual operation scene.

Images