An AGV path planning method based on a flexible spatiotemporal network model

Abstract

The present application provides an AGV path planning method based on a flexible spatiotemporal network model. The present application determines the sub-target tasks of each automated guided vehicle by obtaining the target task; determines the decision variables and their spatiotemporal constraints based on the target task, obtains the time target function and the energy consumption target function for model construction, and introduces a hybrid meta-heuristic algorithm to obtain the AGV spatiotemporal network model; inputs the information of each automated guided vehicle and its corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and driving time of each automated guided vehicle, that is, constructs a hybrid heuristic algorithm based on a distribution estimation algorithm and a genetic algorithm to solve the model, obtains the path planning result of the task vehicle and the driving time of the vehicle on each track by decoding the approximate optimal solution, ensures that there is no collision between the AGVs, and reduces the total energy consumption of the vehicle while ensuring that the target task completion time is as short as possible.

Description

Technical Field

[0001] The present application relates to the field of computers, and in particular to an Automatic Guided Vehicle (AGV) path planning method based on a flexible spatiotemporal network model. Background Art

[0002] In existing technology, AGVs are computer-controlled, driverless vehicles used to transport materials. Since their introduction in 1955, automated guided vehicles (AGVs) have been successfully applied in a variety of fields, including warehouses, container terminals, transportation, and manufacturing systems. In recent years, with the rapid development of intelligent manufacturing, AGVs have become increasingly popular among manufacturers due to their simple operation, quick response, and high efficiency. The primary purpose of using AGV systems is to improve productivity and reduce costs. To achieve this, the control system must effectively plan the AGV's path, namely, coordinate the vehicles to avoid conflicts, collisions, and deadlocks.

[0003] The AGV routing problem can be divided into two consecutive phases: the scheduling phase and the path planning phase. In the scheduling phase, tasks are assigned to AGVs, and then the tasks assigned to each AGV are ranked. Based on the sequence generated in the scheduling phase, the optimal path is found in the path planning phase. Numerous studies have been conducted on this topic by scholars both domestically and internationally. Under the assumption that vehicle speed is fixed and cannot be adjusted, the Dispatching and Conflict-Free Routing Problem (DCFRP) with time windows has been studied in detail. To optimize task allocation, objective functions such as completion time and total completion time are used. In the field of AGVs used in manufacturing systems, the energy consumption of mobile robots is also a major concern for AGV-based systems. However, literature on this topic is limited. Reducing the system's total energy consumption while minimizing task completion time is a key research area for researchers in this field. Summary of the Invention

[0004] One purpose of this application is to provide an AGV path planning method based on a flexible spatiotemporal network model to solve the problem in the prior art of how to ensure task completion time, reasonably optimize the path, and reduce total energy consumption.

[0005] According to one aspect of the present application, a method for AGV path planning based on a flexible spatiotemporal network model is provided, comprising:

[0006] Obtaining a target task, and determining a sub-target task for each automated guided vehicle based on the target task;

[0007] Based on the target task, the decision variables and their spatiotemporal constraints are determined, the time objective function and the energy consumption objective function are obtained to construct the model, and a hybrid meta-heuristic algorithm is introduced to obtain the AGV spatiotemporal network model;

[0008] The information of each automated guided vehicle and the corresponding sub-target task are input into the AGV spatiotemporal network model to obtain the path planning result and driving time of each automated guided vehicle.

[0009] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the decision variables and their spatiotemporal constraints are determined based on the target task, and the time objective function and energy consumption objective function are obtained to construct the model, including:

[0010] The target task includes the task starting point, task end point, cargo information, start time, and planning time. The planning time is discretized into several time intervals, and a mathematical description is performed based on the target task to obtain relevant decision variables.

[0011] respectively performing time constraints, space constraints, and spatiotemporal constraints on the decision variables;

[0012] The time objective function and the energy consumption objective function are determined based on the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints.

[0013] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the time objective function and the energy consumption objective function are determined based on the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints, including:

[0014] A time objective function is obtained by using the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints to determine the total travel time problem and based on mixed integer optimization;

[0015] Determining a total energy consumption problem based on the total driving time problem, wherein the total energy consumption in the total energy consumption problem includes vehicle acceleration energy consumption and rolling friction energy consumption;

[0016] The acceleration variable and its constraint and the time slot number variable are added to the total energy consumption problem, and the energy consumption objective function is obtained based on mixed integer nonlinear programming.

[0017] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the hybrid metaheuristic algorithm is introduced to obtain the AGV spatiotemporal network model, including:

[0018] A hybrid heuristic algorithm two-dimensional coding framework is designed for the driving time and the path planning result, a population incremental learning algorithm is used to construct an AGV probability model, and a penalty function and the energy consumption objective function are used to form a fitness function to obtain an AGV spatiotemporal network model.

[0019] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the first dimension in the two-dimensional coding framework of the hybrid heuristic algorithm corresponds to the path planning result, and the second dimension corresponds to the driving time.

[0020] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the AGV probability model is constructed based on the time dimension using the population incremental learning algorithm, and the AGV probability model includes a time population, an energy consumption population and a learning population, and the time population, the energy consumption population and the learning population are set according to a preset ratio.

[0021] Furthermore, in the above-mentioned AGV path planning method based on the flexible spatiotemporal network model, the information of each automated guided vehicle and its corresponding sub-target task are input into the AGV spatiotemporal network model to obtain the path planning result and travel time of each automated guided vehicle, including:

[0022] Inputting the information of each automated guided vehicle and the corresponding sub-target task into the AGV spatiotemporal network model;

[0023] Based on the information of each automated guided vehicle and the corresponding sub-target task, an initial population is obtained through the AGV probability model;

[0024] Based on the initial population, the fitness function is used to calculate the elite population, and a mutation operation is performed to generate the path vector of the new generation population. The time vector of the new generation population is obtained through the AGV probability model according to the elite population. The above steps are repeated until a preset number of fitness evaluation calculations are completed, and the optimal path vector and time vector of the new generation population are obtained and output as the path planning result and travel time of each automated guided vehicle.

[0025] Compared with the prior art, the present application obtains the target task, determines the sub-target task of each automated guided vehicle based on the target task; determines the decision variables and their spatiotemporal constraints based on the target task, obtains the time target function and the energy consumption target function for model construction, and introduces a hybrid meta-heuristic algorithm to obtain the AGV spatiotemporal network model; inputs the information of each automated guided vehicle and its corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and driving time of each automated guided vehicle, that is, constructs a hybrid heuristic algorithm based on the distribution estimation algorithm and the genetic algorithm to solve the model, obtains the path planning result of the vehicle with a known target task and the driving time of the vehicle on each track by decoding the approximate optimal solution, ensures that there is no collision between the AGVs, and reduces the total energy consumption of the vehicle while ensuring that the target task completion time is as short as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0027] Figure 1 A schematic diagram illustrating a flow chart of an AGV path planning method based on a flexible spatiotemporal network model according to one aspect of the present application is shown;

[0028] Figure 2 A schematic diagram illustrating cumulative flow variables a(i, j, k, t) and d(i, j, k, t) in an embodiment of an AGV path planning method based on a flexible spatiotemporal network model according to one aspect of the present application;

[0029] Figure 3 A schematic diagram of a hierarchical architecture of an AGV system is shown in an embodiment of an AGV path planning method based on a flexible spatiotemporal network model according to one aspect of the present application.

[0030] The same or similar reference numerals in the drawings represent the same or similar components. DETAILED DESCRIPTION

[0031] The present application is described in further detail below with reference to the accompanying drawings.

[0032] Figure 1 A module schematic diagram of an AGV path planning method based on a flexible spatiotemporal network model according to one aspect of the present application is shown. The method is applicable to path planning of unmanned vehicles in industrial manufacturing processes such as warehouses and manufacturing workshops. The method includes steps S1, S2, and S3, which specifically include:

[0033] Step S1, obtaining a target task, and determining a sub-target task for each automated guided vehicle based on the target task;

[0034] Step S2: determining decision variables and their spatiotemporal constraints based on the target task, obtaining a time objective function and an energy consumption objective function for model construction, and introducing a hybrid metaheuristic algorithm to obtain an AGV spatiotemporal network model;

[0035] In step S3, the information of each AGV and its corresponding sub-target task is input into the AGV spatiotemporal network model to obtain the path planning results and travel time for each AGV. In this way, in actual industrial manufacturing processes, intelligently controlling AGVs for cargo transfer reduces labor costs, improves productivity, coordinates multiple vehicles to complete target tasks, avoids conflicts, collisions, and deadlocks, and effectively controls energy consumption through rationally planning routes and travel times.

[0036] In the above steps S1 to S3, the present application obtains the target task, determines the sub-target task of each automated guided vehicle based on the target task; determines the decision variables and their spatiotemporal constraints based on the target task, obtains the time target function and the energy consumption target function for model construction, and introduces a hybrid meta-heuristic algorithm to obtain the AGV spatiotemporal network model; inputs the information of each automated guided vehicle and its corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and driving time of each automated guided vehicle, that is, constructs a hybrid heuristic algorithm based on the distribution estimation algorithm and the genetic algorithm to solve the model, obtains the path planning result of the vehicle with a known target task and the driving time of the vehicle on each track by decoding the approximate optimal solution, ensures that there is no collision between the AGVs, and reduces the total energy consumption of the vehicle while ensuring that the target task completion time is as short as possible.

[0037] Following the above embodiment, in the AGV path planning method of the flexible spatiotemporal network model, the decision variables and their spatiotemporal constraints are determined based on the target task, and the time objective function and the energy consumption objective function are obtained to construct the model, including:

[0038] The target task includes the task starting point, task end point, cargo information, start time, and planning time. The planning time is discretized into several time intervals, and a mathematical description is performed based on the target task to obtain relevant decision variables.

[0039] respectively performing time constraints, space constraints, and spatiotemporal constraints on the decision variables;

[0040] The time objective function and the energy consumption objective function are determined based on the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints. The method facilitates further obtaining the time objective function and the energy consumption objective function by determining decision variables related to travel time and energy consumption and performing constraint-guaranteed collision-free path planning between automated guided vehicles.

[0041] For example, in manufacturing or warehouse systems, AGVs are often used to transport materials between the starting point and destination of a task in a mesh routing environment. In a mesh routing environment, the map is represented as a directed graph G = (V, E), where V is a set of nodes representing loading and unloading locations and locations where the AGV can change direction; E = {(i, j) | i∈V, j∈V, i≠j} is an arc set representing the connection path between two adjacent nodes. Each task of each vehicle is to transport materials from its starting point to its destination. Regarding this process, the following assumptions are made:

[0042] Assume that the size of the AGV is small enough relative to the AGV as a virtual point;

[0043] Each AGV can transport a single load at any time;

[0044] A node can only be occupied by one AGV at a time;

[0045] An arc can be occupied by at most one AGV at a time;

[0046] Turning time is included in the transportation time, and the time for loading and unloading cargo is negligible compared to the transportation time;

[0047] Given the number of vehicles, the transportation task assigned to each AGV is determined, and there are no repeated nodes in the assigned tasks.

[0048] Based on the above assumptions, an AGV spatiotemporal network model is established. The planning time T×Δt is discretized into a series of time slots {Δt, 2Δt, ..., T×Δt}, and the entire path process of the AGV is divided into T time slots for processing.

[0049] First, the symbols and corresponding descriptions of the model are introduced, as shown in Table 1.

[0050] Table 1. Model symbols and descriptions

[0051]

[0052] The decision variables of this system model are as follows:

[0053] x(i, j, k): 0-1 variable, 1 if arc (i, j) is selected as part of the path of AGV k, 0 otherwise;

[0054] y(i, j, k, t): 0-1 variable, 1 if the AGV k occupies arc (i, j) in the tth time slot, 0 otherwise;

[0055] a(i, j, k, t): 0-1 variable, 1 if the AGV k has reached arc (i, j) in the tth time slot, 0 otherwise;

[0056] d(i, j, k, t): 0-1 variable, 1 if the AGV k has left the arc (i, j) at the tth time interval, 0 otherwise;

[0057] TT(i, j, k): integer variable, representing the running time of automatic guided vehicle k in arc (i, j).

[0058] Among them, such as Figure 2 FIG. 1 is a schematic diagram of cumulative flow variables a(i, j, k, t) and d(i, j, k, t) in one embodiment. Figure 2 It can be seen that the AGV k arrives at arc (i, j) at time 4 and leaves arc (i, j) at time 7. The number of time windows in which the AGV k runs on arc (i, j) can be expressed as ∑ t a(i, j, k, t)-∑ t d(i, j, k, t), that is, 3 time windows.

[0059] The flexible spatiotemporal network model implements collision-free guidance paths between AGVs through constraints between variables. Constraints include time constraints, spatial constraints, and spatiotemporal constraints, as follows:

[0060] Space constraints:

[0061]

[0062]

[0063]

[0064] Space-time constraints:

[0065]

[0066]

[0067]

[0068]

[0069]

[0070] Time constraints:

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] Constraints (1)-(3) correspond to the spatial constraints of the AGV's path starting point, intermediate point, and end point, respectively. Constraint (4) is the constraint between the time when vehicle k arrives at arc (i, j) and the time when it leaves arc (i, j). Constraint (5) ensures the continuity of vehicle k in time at point j. Constraint (6) links the time and space constraints together. Constraint (7) describes the variable y(i, j, k, t), which indicates whether arc (i, j) is occupied by vehicle k at time t. Constraint (8) ensures that an arc can be occupied by at most one AGV at the same time. Constraint (9) describes the variable TT(i, j, k), which indicates the running time of vehicle k on arc (i, j). Constraint (10) is the minimum running time (travel time) constraint. Constraint (11) indicates that at most one AGV can occupy each station at the same time. Constraints (12) and (13) ensure temporal continuity.

[0077] The time objective function and the energy consumption objective function are determined based on the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints.

[0078] Following the above embodiment, in the AGV path planning method of the flexible spatiotemporal network model, determining the time objective function and the energy consumption objective function based on the decision variables and their corresponding time constraints, spatial constraints, and spatiotemporal constraints includes:

[0079] A time objective function is obtained by using the decision variables and their corresponding time constraints, space constraints, and spatiotemporal constraints to determine the total travel time problem and based on mixed integer optimization;

[0080] Determining a total energy consumption problem based on the total driving time problem, wherein the total energy consumption in the total energy consumption problem includes vehicle acceleration energy consumption and rolling friction energy consumption;

[0081] The acceleration variable, its constraints, and the number of time slots are added to the total energy consumption problem, and an energy consumption objective function is derived based on mixed-integer nonlinear programming. A model is constructed based on this energy consumption objective function to ensure that each automated guided vehicle achieves low energy consumption after rationally planned paths, thereby achieving the target mission while saving costs.

[0082] For example, the AGV path planning problem includes the total travel time problem and the total energy consumption problem. The primary goal is to optimize the total travel time of the task, which can be expressed as J1:

[0083] J1=∑ k ∑ t t×∑ i (d(i,j,k,t)-d(i,j,k,t-1))

[0084] The optimization of the objective J1 is a mixed integer optimization, and the solution to this objective is defined as problem P1. P1 can be solved by commercial solvers such as Gurobi.

[0085] Based on problem P1, we further consider the system energy consumption and record it as J2, which mainly consists of two parts. a Energy consumption generated by vehicle acceleration, J r The energy consumption caused by rolling friction is called rolling friction energy.

[0086] J2=J a +J r

[0087] J a =0.5M∑ k ∑ t [v 2 (k, t)-v 2 (k, t-1)], v(k, t)≥v(k, t-1)

[0088] Because J a To include conditional constraints, we need to add a new variable z(k, t): a 0-1 variable that is 1 if the AGV k is accelerating at time t and 0 otherwise. max ≤v(k, t)-v(k, t-1)≤v max , the constraint of z(k, t) can be expressed as:

[0089] v(k, t-1)-v(k, t)≤v max (1-z(k, t))

[0090] v(k,t-1)-v(k,t)≥ε-z(k,t)(v max +ε)

[0091] Where ε is a small tolerance, that is, machine precision.

[0092] v(k, t) can be represented by S(i, j) / TT(i, j, k). However, since TT(i, j, k) may be 0, further processing is required. Here, a discrete integer μ (μ∈{1, 2, ..., h}) is introduced, where h is the maximum number of time slots that the AGV can run on the arc. The variable L(i, j, k, μ) is introduced, indicating that the number of time slots that vehicle k can run on arc (i, j) is μ. Then v(k, t) can be expressed as:

[0093]

[0094] Substitute v(k, t) into J a In the equation, we can get the objective function J2, which is nonlinear. The optimization of the objective function J2 under the minimum condition of J1J1 is denoted as problem P2, which is a mixed integer nonlinear programming problem and can be solved using commercial solvers such as Baron. The hierarchical architecture of the AGV system is as follows: Figure 3 shown.

[0095] The controller of each AGV calculates the shortest time T to run on arc (i, j) min (i, j, k) and passes it to the supervisory controller. The supervisory controller then determines the optimal transport time TT(i, j, k) and passes it to the controller of each AGV, thereby controlling the speed of the AGV.

[0096] Following the above embodiment, the method introduces a hybrid meta-heuristic algorithm to obtain an AGV spatiotemporal network model, including:

[0097] A hybrid heuristic algorithm two-dimensional coding framework for the driving time and the path planning results is designed, a population incremental learning algorithm is used to establish an AGV probability model, and a penalty function and the energy consumption objective function are used to form a fitness function to obtain an AGV spatiotemporal network model. Here, the problem is encoded, which is highly correlated with the quality of the solution. Constructing the two-dimensional coding framework is conducive to improving the quality of the solution; the AGV probability model is related to the performance of the algorithm. Using a population incremental learning algorithm to establish the AGV probability model is conducive to obtaining a good initial solution, thereby improving the performance of the algorithm; introducing a fitness function ensures that high-quality solutions have good fitness and avoids the algorithm from searching in an infeasible space. For example, the fitness function F(X) of solution X consists of two parts: the objective function J(X) = J2 and the penalty function p(X). The penalty function p(X) prevents the algorithm from searching in an infeasible solution space, as follows:

[0098] p(X)=p1(X)+p2(X)

[0099]

[0100]

[0101] in, is the shortest completion time obtained by solving the traditional spatiotemporal network model. R1 and R2 are both relatively large constants. p1 ensures that solutions with shorter completion times have lower fitness values, while p2 ensures that infeasible solutions have lower fitness values.

[0102] Following the above embodiment of the present application, the first dimension of the two-dimensional coding framework of the hybrid heuristic algorithm in the method corresponds to the path planning result, and the second dimension corresponds to the driving time. Here, the first dimension corresponds to the path planning result of the solution, which is solved by GA, and the second dimension corresponds to the driving time of the solution, which is solved by EDA. In the designed framework, each individual contains two parts of information: path vector With time vector in, is a variable representing the site selected at the lth position of the path vector of vehicle k, is a variable representing the time slot selected by the lth position of the time vector of vehicle k. And Variables corresponding to the spatiotemporal network model Thus, if For l∈{N1,...,N}

[0103] Following the above embodiment of the present application, the above method adopts the population incremental learning algorithm based on the time dimension to construct the AGV probability model, and the AGV probability model includes a time population, an energy consumption population and a learning population, and the time population, the energy consumption population and the learning population are set according to a preset ratio.

[0104] For example, for the time dimension, each AGV has N-1 variables that need to be determined, and the group incremental learning algorithm is used here to build a probability model. The probability distribution model can be described as:

[0105]

[0106] The time probability matrix C of vehicle k at the iter-th iteration k (iter) is designed as follows:

[0107]

[0108] in, Represents probability According to the dual objective function of the research problem P2, the algorithm population is divided into three parts: time population P t , energy consumption population P e And the learning population P l , the ratio of the three populations is set to 1:1:2. The probability matrix is ​​initialized as follows:

[0109] For the time population P t AGV runs at the highest speed to minimize the total completion time, time probability matrix The initial setting is as shown in the above formula:

[0110]

[0111] For the energy consumption population P e AGV travels at a relatively high speed to reduce energy consumption, energy consumption probability matrix The initial setting is as follows:

[0112]

[0113] Learning population P l The probability matrix is ​​initialized as shown in the following formula. The probability of each time slot being selected is the same, which is 1 / h.

[0114]

[0115] In another embodiment of the present application, the method of inputting the information of each automated guided vehicle and its corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and travel time of each automated guided vehicle includes:

[0116] Inputting the information of each automated guided vehicle and the corresponding sub-target task into the AGV spatiotemporal network model;

[0117] Based on the information of each automated guided vehicle and the corresponding sub-target task, an initial population is obtained through the AGV probability model;

[0118] Based on the initial population, the fitness function is used to calculate the elite population, and a mutation operation is performed to generate the path vector of the new generation population. The time vector of the new generation population is obtained through the AGV probability model according to the elite population. The above steps are repeated until a preset number of fitness evaluation calculations are completed, and the optimal path vector and time vector of the new generation population are obtained and output as the path planning result and travel time of each automated guided vehicle.

[0119] For example, in step 1, the initial population is first determined according to the method for generating the initial solution, that is, the initial population is obtained through the AGV probability model, and the initial population is divided into three categories.

[0120] Step 2: Based on the initial group, the fitness function is used to calculate the population fitness, and the elite population is selected according to the fitness.

[0121] Step 3: Genetic Algorithm is used to perform mutation operations to generate the path vectors of the new generation population. The probability of selecting the three mutation operations (flip, swap, and slide) is the same. Dijkstra algorithm is used to ensure the continuity of the path.

[0122] Step 4: Update the probability matrix C according to the elite population k(iter), using the probability update method of the Population-Based Incremental Learning (PBIL) algorithm.

[0123] Step 5: Obtain the time vector of the new generation population according to probability matrix sampling.

[0124] Step 6: Determine whether the fitness evaluation calculation has been performed for a preset number of times. If satisfied, output the result; otherwise, return to step 2.

[0125] The specific hybrid heuristic algorithm pseudo code is as follows:

[0126] Input: AGV task allocation

[0127] Output: optimal individual X best

[0128]

[0129]

[0130] In summary, the present application obtains the target task, determines the sub-target task of each automated guided vehicle based on the target task; determines the decision variables and their spatiotemporal constraints based on the target task, obtains the time target function and the energy consumption target function for model construction, and introduces a hybrid meta-heuristic algorithm to obtain the AGV spatiotemporal network model; inputs the information of each automated guided vehicle and its corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and driving time of each automated guided vehicle, that is, constructs a hybrid heuristic algorithm based on the distribution estimation algorithm and the genetic algorithm to solve the model, obtains the path planning result of the vehicle with a known target task and the driving time of the vehicle on each track by decoding the approximate optimal solution, ensures that there is no collision between the AGVs, and reduces the total energy consumption of the vehicle while ensuring that the target task completion time is as short as possible.

[0131] It should be noted that the application can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In one embodiment, the software program of the application can be executed by a processor to realize the steps or functions described above. Similarly, the software program of the application (including relevant data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and similar devices. In addition, some steps or functions of the application can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0132] In addition, a part of the present application may be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can call or provide the method and / or technical solution according to the present application through the operation of the computer. The program instructions for calling the method of the present application may be stored in a fixed or removable recording medium, and / or transmitted through a data stream in a broadcast or other signal-carrying medium, and / or stored in a working memory of a computer device that runs according to the program instructions. Here, according to an embodiment of the present application, a device is included, which includes a memory for storing computer program instructions and a processor for executing program instructions, wherein, when the computer program instructions are executed by the processor, the device is triggered to run the method and / or technical solution based on the aforementioned multiple embodiments of the present application.

[0133] It is obvious to those skilled in the art that the present application is not limited to the details of the above-mentioned exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or basic characteristics of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. Words such as first and second are used to indicate names and do not indicate any particular order.

Claims

1. An AGV path planning method based on a flexible spatiotemporal network model, characterized in that: The method comprises: Obtaining a target task, and determining a sub-target task for each automated guided vehicle based on the target task; Based on the target task, the decision variables and their spatiotemporal constraints are determined, and the time objective function and energy consumption objective function are obtained to construct the model. A hybrid metaheuristic algorithm combining the genetic algorithm and the distribution estimation algorithm is introduced to obtain the AGV spatiotemporal network model. The step of determining the decision variables and their spatiotemporal constraints based on the target task, obtaining the time objective function and the energy consumption objective function for model construction includes: The target task includes the task starting point, task end point, cargo information, start time, and planning time. The planning time is discretized into a number of time intervals, and a mathematical description is performed based on the target task to obtain relevant decision variables. The decision variables are respectively subject to time constraints, space constraints, and spatiotemporal constraints. The time objective function and the energy consumption objective function are determined based on the decision variables and their corresponding time constraints, the space constraints, and the spatiotemporal constraints, including: determining a total driving time problem using the decision variables and their corresponding time constraints, the space constraints, and the spatiotemporal constraints, and obtaining a time objective function based on mixed integer optimization; determining a total energy consumption problem based on the total driving time problem, wherein the total energy consumption in the total energy consumption problem includes vehicle acceleration energy consumption and rolling friction energy consumption; adding an acceleration variable and its constraints and a time gap number variable to the total energy consumption problem, and obtaining an energy consumption objective function based on mixed integer nonlinear programming. The specific formula is as follows: J2=J a +J r J a =0.5MΣ k Σ t [v 2 (k,t)-v 2 (k,t-1)],v(k,t)≥v(k,t-1) J2 is the total energy consumption, J a is the vehicle acceleration energy consumption, J r is the rolling friction energy consumption, v is the speed of the automatic guided vehicle, and since J a To include conditional constraints, a new variable z(k,t) needs to be added: a 0-1 variable, which is 1 when the AGV k accelerates at time t, and 0 otherwise; max ≤v(k, t)-v(k, t-1)≤v max , the constraint of z(k, t) can be expressed as: v(k,t-1)-v(k,t)≤v max (1-z(k,t)) v(k,t-1)-v(k,t)≥ε-z(k,t)(v max +s) Where ε is a smaller tolerance, that is, the machine precision; v(k, t) can be represented by S(i, j) / TT(i, j, k), where TT(i, j, k) represents the running time of the automated guided vehicle k on arc (i, j), and S(i, j) represents the path length of arc (i, j). However, since TT(i, j, k) may be 0, further processing is required. A discrete integer μ(μ∈{1, 2, ..., h}) is introduced, where h is the maximum number of time slots that the AGV can run on the arc. The variable L(i, j, k, μ) is introduced, indicating that the number of time slots that vehicle k can run on arc (i, j) is μ. Then v(k, t) can be expressed as: Substitute v(k, t) into J a In the equation, a nonlinear energy consumption objective function J2 can be obtained, and the optimization of the energy consumption objective function J2 under the condition that the time objective function J1 is minimized is recorded as the total energy consumption problem P2; The information of each AGV and its corresponding sub-goal task are input into the AGV spatiotemporal network model to obtain the path planning result and travel time of each AGV, wherein the controller of each AGV calculates the shortest time T running on arc (i, j) min (i, j, k) and pass it to the supervisory controller, which determines the optimal transportation time TT(i, j, k) and passes it to the controller of each AGV, thereby controlling the speed of the AGV.

2. The method according to claim 1, characterized in that The hybrid meta-heuristic algorithm is introduced to obtain the AGV spatiotemporal network model, including: A hybrid heuristic algorithm two-dimensional coding framework is designed for the driving time and the path planning result, a population incremental learning algorithm is used to establish an AGV probability model, and a penalty function and the energy consumption objective function are used to form a fitness function to obtain an AGV spatiotemporal network model.

3. The method according to claim 2, characterized in that The first dimension in the two-dimensional coding framework of the hybrid heuristic algorithm corresponds to the path planning result, and the second dimension corresponds to the driving time.

4. The method according to claim 3, characterized in that The AGV probability model is constructed based on the time dimension using the population incremental learning algorithm. The AGV probability model includes a time population, an energy consumption population, and a learning population, and the time population, the energy consumption population, and the learning population are set according to a preset ratio.

5. The method according to any one of claims 1 to 4, characterized in that Inputting the information of each automated guided vehicle and the corresponding sub-target task into the AGV spatiotemporal network model to obtain the path planning result and travel time of each automated guided vehicle includes: Inputting the information of each automated guided vehicle and the corresponding sub-target task into the AGV spatiotemporal network model; Based on the information of each automated guided vehicle and the corresponding sub-target task, an initial population is obtained through the AGV probability model; Based on the initial population, an elite population is calculated using a fitness function, and a mutation operation is performed to generate a path vector of a new generation population. The time vector of the new generation population is obtained through the AGV probability model according to the elite population. The above steps are repeated until a preset number of fitness evaluation calculations are completed, and the optimal path vector and time vector of the new generation population are obtained and output as the path planning result and travel time of each automated guided vehicle.

Images