Path planning method and device for automatic conveying device

Abstract

The invention provides a path planning method and device for an automatic conveying device. The path planning method comprises the following steps: determining an initial position and a target position of the automatic conveying device in a target environment; searching a passing position of a path from the initial position to the target position according to a target search direction by taking the initial position as a starting point; planning a possible path from the starting position to the target position based on the passing position; from the possible paths, a target path for the automated transport device from the starting position to the target position is determined. According to the method, the problems of large calculation amount, long calculation time and difficulty in improving the path planning efficiency in path planning are solved, the path position can be searched according to the target search direction, all position nodes and paths do not need to be traversed and checked, and the path planning process is simplified, so that the search range can be narrowed, and the calculation speed and efficiency of path planning are improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of automatic path planning, and more specifically, to a path planning method and device for an automatic transport device. Background Art

[0002] With the development of automation technology, automatic transportation devices such as automated guided vehicles (AGVs) can automatically transport the materials they carry to designated locations according to the needs of production and life, thereby greatly saving the cost and time of manual transportation.

[0003] For such an automatic transport device, path planning can be performed based on the starting and end points of the transport task. On the one hand, it is necessary to plan a reachable path to complete the transport; on the other hand, it is also necessary to optimize the path planning process, improve the planning calculation speed, and find a better path.

[0004] In the path planning schemes of related technologies, it is usually necessary to check all nodes and paths by traversing, and in some cases, backpropagation is required to update the planned path. Such schemes often require complex calculation processes and large amounts of calculation, making it difficult to improve the efficiency of path planning. In particular, in path planning scenarios with large site areas, the problem of time-consuming calculations will be more prominent. Summary of the invention

[0005] In view of the problems that the path planning schemes in the related art have large amount of calculation, long calculation time, and difficulty in improving the efficiency of path planning, the present disclosure provides a path planning method and device for an automatic transport device.

[0006] A first aspect of the present disclosure provides a path planning method for an automatic transport device, the path planning method comprising: determining a starting position and a target position of the automatic transport device in a target environment; taking the starting position as a starting point, searching for a path from the starting position to the target position through a pass position according to a target search direction, wherein the target search direction is determined based on a positional relationship between a reference position and the target position; searching for a path from the starting position to the target position through a pass position according to the target search direction; planning a possible path from the starting position to the target position based on the pass position; and determining a target path of the automatic transport device from the starting position to the target position from the possible paths.

[0007] Optionally, the target environment is an assembly environment of a wind turbine generator set, and the target search direction includes: a direction pointing from the reference position to the target position, wherein the passing position is searched in the following manner: taking the starting position as the starting point, searching for the passing position that satisfies the target search direction, wherein satisfying the target search direction means: the direction pointing from the reference position to the passing position is the same as the target search direction or the same as a component of the target search direction.

[0008] Optionally, the search for the via position that satisfies the target search direction starting from the starting position includes: performing a search operation on the optional positions in the target environment until the target position is searched, wherein the search operation is performed in the following manner: determining a first position that satisfies an adjacency condition with the current reference position from the optional positions in the target environment, wherein the adjacency condition includes: the first position is adjacent to the current reference position, and there is no obstacle between the first position and the current reference position; determining a second position that satisfies a direction condition with the current reference position from the first position, wherein the direction condition includes: the direction from the current reference position to the second position is the same as the current target search direction or the same as a component of the current target search direction; in response to the second position being different from the target position, determining the second position as the via position, and using the second position as the reference position for the next search operation, and performing the next search operation; in response to the second position being the same as the target position, ending the search operation, wherein the reference position for the first search operation is the starting position.

[0009] Optionally, there are multiple possible paths, and determining the target path of the automatic transport device from the starting position to the target position from the possible paths includes: determining a path coefficient for each possible path, wherein the path coefficient represents: the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; and determining the possible path with the smallest path coefficient as the target path.

[0010] Optionally, the automatic transport device is determined to have experienced a turn by: in response to an angular difference between a direction of travel of the automatic transport device when moving to a current position and a direction of travel of the automatic transport device when moving to a next position, determining that the automatic transport device has experienced a turn when moving from a current position to a next position.

[0011] Optionally, the target environment is an assembly environment of a wind turbine generator set, wherein determining the path coefficient of each possible path includes: determining a distance coefficient and a turning coefficient respectively according to the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; determining a turning weight according to the area size of the assembly environment, wherein the turning weight is positively correlated with the area size; weighting the turning coefficient using the turning weight to obtain a weighted turning coefficient; and obtaining the path coefficient based on the distance coefficient and the weighted turning coefficient.

[0012] Optionally, searching for the passing positions along the path from the starting position to the target position according to the target search direction includes: dividing a target area in the target environment according to the starting position, the target position and an environmental boundary of the target environment, wherein the target area includes the starting position and the target position; and searching for the passing positions in the target area according to the target search direction.

[0013] Optionally, the target area is determined in the following manner: determining a first candidate position between the starting position and the target position that is closest to the environment boundary in a preset direction; determining a second candidate position along the preset direction that is a preset distance away from the first candidate position; and determining a boundary of the target area in the preset direction based on the positional relationship between the second candidate position and the environment boundary to determine the target area.

[0014] Optionally, there are multiple automatic conveying devices, each of which corresponds to its own target position, wherein the path planning method further includes: determining the initial path of each automatic conveying device according to the number of target paths corresponding to each automatic conveying device; in response to the presence of conflicting path segments between the initial paths of each automatic conveying device, updating the initial paths of the conflicting automatic conveying devices by performing a path update operation until there is no conflict between the initial paths of each automatic conveying device, thereby obtaining the final paths of each automatic conveying device, wherein the conflict between the target paths of the automatic conveying devices means that the automatic conveying devices will be in the same path segment at the same time when traveling along their respective target paths; in response to the absence of conflicting path segments between the initial paths of each automatic conveying device, taking the current initial path of each automatic conveying device as the final path of each automatic conveying device.

[0015] Optionally, the initial path of the conflicting automatic transport device is updated in the following manner: by adjusting the time when at least one of the conflicting target transport devices reaches the conflicting path segment, the initial path of the at least one target transport device is updated.

[0016] Optionally, the target environment is an assembly environment of a wind turbine generator set, wherein the initial path of the at least one target transport device is updated in the following manner: determining a distance index between the starting position and the target position of each of the conflicting target transport devices, wherein the distance index is used to characterize the distance between the starting position and the target position; for any two conflicting target transport devices, determining the initial paths of the two conflicting target transport devices by performing the following operations: in response to a target position of a first target transport device being located on the initial path of a second target transport device, determining a waiting time for the second target transport device before entering the conflicting path segment based on a length of a conflicting path segment between the first target transport device and the second target transport device, a moving speed of the first target transport device, and a preset stay time, so as to update the initial path of the second target transport device; in response to a target position of the first target transport device not being located on the initial path of the second target transport device, determining a waiting time for the second target transport device before entering the conflicting path segment based on a length of a conflicting path segment between the first target transport device and the second target transport device and a moving speed of the first target transport device, so as to update the initial path of the second target transport device, wherein the distance index of the first target transport device is greater than the distance index of the second target transport device.

[0017] Optionally, the initial path of each automatic conveying device is determined in the following manner: in response to the number of target paths of the first automatic conveying device being less than the number of target paths of the second automatic conveying device, the target path among the target paths of the second automatic conveying device that does not overlap with the initial path of the first automatic conveying device is used as the initial path of the second automatic conveying device; in response to the number of target paths of the first automatic conveying device being equal to the number of target paths of the second automatic conveying device, the initial paths of the first automatic conveying device and the second automatic conveying device are determined from their respective target paths, so that the initial paths of the first automatic conveying device and the second automatic conveying device do not overlap.

[0018] A second aspect of the present disclosure provides a path planning device for an automatic transport device, the path planning device comprising: a position determination unit, configured to determine a starting position and a target position of the automatic transport device in a target environment; a search unit, configured to take the starting position as a starting point and search for a path passing through a path from the starting position to the target position according to a target search direction, wherein the target search direction is determined based on a positional relationship between a reference position and the target position; a planning unit, configured to plan a possible path from the starting position to the target position based on the passing positions; and a path determination unit, configured to determine a target path of the automatic transport device from the starting position to the target position from the possible paths.

[0019] According to the path planning method and device of the automatic transport device disclosed in the present invention, the target search direction determined based on the positional relationship between the reference position of the automatic transport device in the target environment and the target position can be used to search for the passing position from the starting position to the target position, and a possible path that can reach the target position can be planned to determine the final target path. In this way, the path position can be searched according to the target search direction without traversing and checking all position nodes and paths, simplifying the path planning process, thereby narrowing the search range and improving the calculation speed and efficiency of path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a schematic flowchart illustrating a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0021] Figure 2 1 is a schematic plan view showing a target environment in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0022] Figure 3 is a plan schematic diagram showing a rasterized target environment in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0023] Figure 4 is a schematic diagram showing an example of a target environment in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0024] Figure 5 is a schematic flowchart illustrating steps of a search operation in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0025] Figure 6 is a schematic diagram illustrating region segmentation in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0026] Figure 7 is a schematic flowchart illustrating steps of determining a target area in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0027] Figure 8 and Fig. 9 They are schematic diagrams showing example results of searching for passing locations in a path planning method of the related art and a path planning method of an automatic transport device according to an exemplary embodiment of the present disclosure, respectively.

[0028] Fig.10 Detailed description is a schematic flowchart illustrating steps of determining a target path in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0029] Fig.11 is a schematic diagram illustrating determining an angle difference in a path planning method for an automatic conveying device according to an exemplary embodiment of the present disclosure.

[0030] Fig.12 is a schematic flowchart illustrating steps of a path updating operation in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0031] Fig.13 1 is a schematic flowchart illustrating steps of a path updating operation when a conflict exists in an initial path in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0032] Fig.14A and Fig. 14B is a schematic flowchart illustrating an application example of a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0033] Fig.15 1 is a plan view schematically showing a transport environment including a plurality of automatic transport devices in a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure.

[0034] Fig.16 It is a plan schematic diagram showing the transport environment shown in the figure after rasterization processing.

[0035] Fig.17A and Fig. 17B It is a comparison curve diagram showing the possible paths of each automatic transport device in the transport environment shown in the figure and the final target path.

[0036] Fig.18 3 is a graph showing a comparison of calculation times of a path planning method for an automatic conveying device according to an exemplary embodiment of the present disclosure and a path planning method of the related art.

[0037] Fig.193 is a graph showing a comparison of energy consumption between a path planning method for an automatic conveying device according to an exemplary embodiment of the present disclosure and a path planning method of the related art.

[0038] Fig. 20 3 is a graph showing a comparison of the path calculation time of the path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure and the path planning method of the related art as the number of nodes increases.

[0039] Fig.21 3 is a graph comparing energy consumption of a path planning method for an automatic transport device according to an exemplary embodiment of the present disclosure and a path planning method in the related art as the number of nodes increases.

[0040] Fig. 22 is a schematic block diagram showing a path planning device of an automatic transport device according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0041] The following specific embodiments are provided to help the reader obtain a comprehensive understanding of the methods, devices and / or systems described herein. However, after understanding the disclosure of the present application, various changes, modifications and equivalents of the methods, devices and / or systems described herein will be clear. For example, the order of operations described herein is only an example and is not limited to those orders set forth herein, but can be changed as will be clear after understanding the disclosure of the present application, except for operations that must occur in a specific order. In addition, for greater clarity and simplicity, the description of features known in the art may be omitted.

[0042] The features described herein can be implemented in different forms and should not be construed as being limited to the examples described herein. Rather, the examples described herein have been provided to illustrate only some of the many possible ways to implement the methods, devices, and / or systems described herein, which will be clear after understanding the disclosure of the present application.

[0043] As used herein, the term "and / or" includes any one of the associated listed items and any combination of any two or more.

[0044] Although terms such as "first", "second", and "third" may be used herein to describe various members, components, regions, layers, or portions, these members, components, regions, layers, or portions should not be limited by these terms. Instead, these terms are only used to distinguish one member, component, region, layer, or portion from another member, component, region, layer, or portion. Therefore, without departing from the teachings of the examples described herein, the first member, first component, first region, first layer, or first portion referred to in the examples may also be referred to as the second member, second component, second region, second layer, or second portion.

[0045] In the specification, when an element (such as a layer, a region, or a substrate) is described as being “on”, “connected to”, or “coupled to” another element, the element may be directly “on”, “connected to”, or “coupled to” another element, or one or more other elements may be present therebetween. Conversely, when an element is described as being “directly on”, “directly connected to”, or “directly coupled to” another element, there may be no other elements present therebetween.

[0046] The terms used herein are only used to describe various examples and are not intended to limit the disclosure. Unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. The terms "comprise", "include" and "have" indicate the presence of the described features, quantities, operations, components, elements and / or combinations thereof, but do not exclude the presence or addition of one or more other features, quantities, operations, components, elements and / or combinations thereof.

[0047] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by a person of ordinary skill in the art to which the present disclosure belongs after understanding the present disclosure. Unless explicitly defined as such herein, terms (such as those defined in a general dictionary) should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and should not be interpreted in an idealized or overly formal manner.

[0048] Furthermore, in the description of examples, when it is considered that a detailed description of a well-known related structure or function would cause vague interpretation of the present disclosure, such a detailed description will be omitted.

[0049] As mentioned above, with the development of automation technology, automatic conveying devices can complete the transportation of loads, save labor costs and improve work efficiency.

[0050] Taking the assembly environment of wind turbines as an example, the assembly efficiency of wind turbines may directly affect the development of the new energy industry. The transportation of materials in the assembly environment of wind turbines is an important part of the assembly work. Among them, AGV path planning in the assembly environment of wind turbines is a typical Industry 4.0 application scenario. With the development of smart manufacturing technology, AGV path planning is not only common in intelligent operation and maintenance, but also extended to wind turbine hoisting, wind farm maintenance, and even the field of the Internet of Things. Multi-AGV path planning has gradually evolved into a multi-objective combinatorial optimization problem: in addition to paying attention to the rapidity of the algorithm, it is also necessary to consider the shortest path distance, minimum time and minimum energy consumption. Therefore, the multi-AGV path planning problem has also received widespread attention.

[0051] Single AGV path planning is the basis of multi-AGV path planning. It belongs to the single-source path optimization problem, and its purpose is to find the shortest path from the specified starting point to the specified target point that satisfies certain constraints. The AGV path planning problem can be solved by graph theory (for example, Dijkstra shortest path algorithm, A* / D* algorithm, probabilistic roadmap, fast random tree), system sampling method (for example, artificial potential field method, neural network) and intelligent algorithm (for example, genetic algorithm, ant colony algorithm). Here, Dijkstra algorithm, A* algorithm and D* algorithm are the three most commonly used path search algorithms. Compared with artificial intelligence technologies such as deep learning, these algorithms have simple structures. However, the essence of these algorithms is breadth-first search. For example, the time complexity of Dijkstr a algorithm and A* algorithm is O(n 2 ); the time complexity of the D* algorithm is O(2 n ), so the computational efficiency is not high in a large environment with multiple nodes.

[0052] Markov Decision Process (MDP) is the modeling basis of dynamic programming and reinforcement learning. With the development of artificial intelligence (AI), MDP has once again set off a research boom. Specifically, dynamic programming is the theoretical basis of reinforcement learning, which can transform a multi-stage optimization process into a series of single-stage optimization problems. However, traditional algorithms iterate the sample set by randomly selecting a starting point until the optimal strategy is obtained. As the scale of the path planning environment expands, the number of training times and the time for a single training session will also increase, so the amount of computation will increase exponentially with the increase in the number of nodes.

[0053] Monte Carlo tree search is an optimization algorithm that builds an optimal action list by constructing an asymmetric tree in a given area, and finds the optimal path from the starting point to the target point by expanding from known nodes to unknown nodes. However, in a large-scale environment with multiple nodes, the traditional Monte Carlo tree search lacks a heuristic mechanism and cannot balance "exploration" and "utilization". In the process of building the search tree, all unexpanded nodes will be traversed one by one, and then the one with the largest upper confidence interval will be selected from all unexpanded nodes as the next target node. This is not applicable in situations where real-time requirements are high. In addition, when the leaf nodes of the search tree are expanded to the target point, the traversed nodes are evaluated, and then the nodes with the highest scores are found among all the traversed nodes to form a path, which also makes it inefficient in large environments.

[0054] In some cases, the target environment, such as the wind turbine assembly environment, may contain multiple relatively regularly arranged workstations. Workstations are nodes that can be used by AGVs. They can be used as locations for storing materials, or they can be left empty as passages for AGVs to travel between workstations to transport materials. If their length and width are ignored, workstations can be abstracted as nodes, so the AGV workflow can be modeled using MDP. However, due to the different layouts of different target environments and the relatively special driving conditions of AGVs, it is impossible to completely replicate the traditional MDP method to model the environment.

[0055] In view of the above-mentioned problems, the present disclosure provides a path planning method and device for an automatic transport device to solve or at least alleviate the above-mentioned problems.

[0056] According to a first aspect of an exemplary embodiment of the present disclosure, a path planning method for an automatic transport device is provided. The path planning method can be executed by a computer device having computing capabilities. The computer device executing the path planning method can be, for example, a terminal device or a server, wherein the terminal device can be, for example, a tablet computer, a laptop computer, a digital assistant, etc.; the server can be an independent server, a server cluster, a cloud computing platform, or a virtualization center.

[0057] Here, the computer device can be set at the automatic transport device or can be communicatively connected to the automatic transport device, so as to control the automatic transport device and obtain the data required to execute the above method, for example, it can remotely send control instructions to the automatic transport device (such as the final planned path), receive the location information of the automatic transport device, etc.

[0058] In an embodiment according to the present disclosure, the path planning method may include the following steps:

[0059] like Figure 1 As shown, in step S110, the starting position and the target position of the automatic transport device in the target environment can be determined.

[0060] Here, the automatic transport device may be, for example, the above-mentioned AGV, but it is not limited thereto, and may also be other devices capable of automatic movement. The target environment may be the transport environment of the automatic transport device, such as, but not limited to, the assembly environment of a wind turbine generator set, etc., which may be determined according to the actual application scenario of the automatic transport device. The embodiments of the present disclosure do not impose any particular restrictions on the automatic transport device and the target environment.

[0061] The starting position may be the current position of the automatic transport device, or may be a position other than the current position of the automatic transport device. For example, when performing a transport task, the automatic transport device may first move to the starting position, and then perform the transport task from the starting position to the target position. Therefore, in some cases, in response to receiving a transport task and the automatic transport device is not currently at the starting position of the transport task, the starting position of the transport task may be used as the current target position, and the automatic transport device may first be moved to the current target position, i.e., the starting position of the transport task, according to the path planning method of an embodiment of the present disclosure; and then, according to the path planning method, the automatic transport device may be moved to the target position specified in the transport task. The target position may be specified in the transport task, and may be predetermined or change in real time.

[0062] In the embodiments of the present disclosure, the starting position and the target position are relative to each other. In different transport tasks or different stages of the same transport task, the current starting position (or target position) can also be used as the target position (or starting position) of another task or stage. In addition, the starting position and the target position can be changed according to the current transport demand or transport stage. For any pair of starting positions and target positions, the path planning method according to the embodiments of the present disclosure can be used to plan the corresponding target path.

[0063] Refer to the following Figure 2 and Figure 3 The target environment and the representation of each position in the target environment are described by taking the assembly environment of a wind turbine as an example.

[0064] Figure 2 The schematic diagram of a wind turbine assembly plant is shown in Figure 2 In the figure, the numbers with circles represent the workstations of the automatic conveyor, among which the numbers in bold represent the charging stations of the automatic conveyor; the numbers without circles represent the intersections of the driving routes. As an example, the gridding method can be used to Figure 2 Rasterize and define coordinates, the result is as follows Figure 3 As shown. Figure 3 The bold numbers correspond to the workstations in Figure 1 The circled numbers indicate the workstations.

[0065] Here, in the rasterization process, the actual environment can be modeled by rasterizing the Cartesian coordinate system. For example, it can be set that in the entire target environment, the position of the upper left corner is (x = 0, y = 0), and the position of the lower right corner is (x = n - 1, y = m - 1), with rows as the x-axis and columns as the y-axis, where n and m are the total number of rows and columns respectively; each grid corresponds to coordinates (a, b), where (a ≤ b - 1, b ≤ m - 1), and the number in the grid represents the number of this node. The target environment can be represented as G(V, O, E), where V represents the set of nodes (V = {v1, v2,..., v p}, p is the total number of nodes). Here, each grid can be used as a node; O represents the set of obstacle nodes (i.e., nodes that the automatic transport device cannot pass through when carrying materials, such as assembly work points) (O = {o1, o2,..., o q}, q is the total number of obstacle nodes, q < p); E represents the set of paths for the automatic transport device to travel. Let the length of the automatic transport device be L AGV and the width be W AGV . The length of each grid (in the x-direction) is L grid = L AGV +δ, and the width (in the y-direction) is W grid = W AGV +δ (where δ is any positive number, which represents the safety margin for the automatic transport device to travel and can be set according to actual requirements). The length of the target environment (in the x-direction) is L WT and the width (in the y-direction) is W WT . Therefore, the number of rows of the target environment is and the number of columns is where [·] represents rounding down. In this way, the target environment can be divided into n grids, n = [R grid ×C grid ′, where [·]′ represents rounding up.

[0066] So far, the rasterization process of the actual target environment is completed, and the positions of each grid can be represented in the form of coordinates, and it corresponds to the position in the actual environment. However, the above-mentioned rasterization and coordinate system establishment methods are only examples, and different positions in the target environment can also be represented in other ways. The embodiments of the present disclosure do not particularly limit the representation methods of the environment and positions.

[0067] In the example of representing each position in the target environment by nodes or grids, the entire target environment can be regarded as composed of a series of regularly arranged nodes or grids, and the immediate rewards obtained by the automatic transport device reaching different nodes can be different.

[0068] For example, the reward definition of the automatic transport device in the target environment can be as follows: (1) If the automatic transport device carrying materials enters an obstacle node (e.g. Figure 3 (1) If the automatic transport device enters an obstacle node without any material, the reward obtained may be 0; (2) If the automatic transport device enters an idle node, the reward obtained may be -δ, where δ is a positive integer and δ < < ε. The automatic transport device receives a penalty δ for each step forward before reaching the target location; (3) If the automatic transport device (empty or loaded) reaches the target location, the reward obtained may be +ε.

[0069] In addition, the reward matrix RM (Reward Matrix) of the target environment can be constructed as follows: For an m×n target environment, m represents the number of column nodes; n represents the number of row nodes. Assume that the starting position of the automatic transport device is (1,1) and the target position is (m,n). Assume that there are 4 obstacle nodes, namely (1,3), (2,2), (2,3) and (3,3), then the reward matrix RM of the target environment can be expressed as the following formula (1):

[0070]

[0071] The dimension of PM matrix is ​​m×n, which is the same as the dimension of the position matrix of the target environment. 11 …r mn They respectively represent the rewards obtained by the automatic transport device when it reaches the node in the 1st row / 1st column to the node in the mth row / nth column in the target environment.

[0072] In addition, except for the nodes on the edge of the target environment, each of the other nodes has four adjacent nodes, but the node located at the upper edge of the target environment has no upper adjacent node, the node located at the left edge has no left adjacent node, the node located at the lower edge has no lower adjacent node, and the node located at the right edge has no right adjacent node. Therefore, the state transition probability matrix PM of the target environment can be constructed as follows: For an m×n target environment, in principle, whether the automatic transport device is loaded or unloaded, the probability of reaching each adjacent node is the same, that is, P ij =P[S t+1 =j|S t =i]=δ, where δ is 0 or 1 (corresponding to the two states of the presence and absence of obstacles, respectively). For example, the state transition probability matrix PM can be expressed as the following formula (2):

[0073]

[0074] In the above representation of the reward matrix and the state transition probability matrix, for a target environment with m×n nodes or grids, the traditional reward matrix has (m×n)×(m×n) rows / (m×n)×(m×n) columns, while the dimensions of the reward matrix and the state transition probability matrix proposed in the embodiments of the present disclosure are both m×n, which are the same as the dimensions of the target environment. In this way, the computational complexity can be reduced and the algorithm efficiency can be improved.

[0075] In step S120, the starting position may be used as a starting point, and according to the target search direction, the passing positions of the path from the starting position to the target position may be searched.

[0076] Here, the target search direction is determined based on the positional relationship between the reference position and the target position. In the first search, the reference position may be the starting position, and in the non-first search, the reference position may be the passing position found in the last search.

[0077] In this step, the positional relationship between the reference position and the target position can represent the direction of the target position relative to the reference position. As an example, the target search direction may include: the direction from the reference position to the target position; and / or, the component of the direction from the reference position to the target position, such as the component direction on the axis of the coordinate system of the target environment.

[0078] by Figure 4 As an example, assuming that the starting position is position 2 and the target position is position 18, the positional relationship between the reference position and the target position during the first search is that the target position 18 is located in the lower right direction relative to the starting position 2. Figure 4 In the example coordinate system of the target environment shown, the target position 18 is located in the positive direction of the x-axis and the positive direction of the y-axis relative to the starting position 8. Accordingly, the target search direction may include: a direction pointing from the current position to the target position; and / or, pointing from the current position to the positive direction of the x-axis and the positive direction of the y-axis.

[0079] Here, the starting position is used as an example for description as the reference position. In the subsequent search process of the passing position, as the search proceeds, the currently searched passing position can be used as a reference position to determine the target search direction of the next search.

[0080] Specifically, in step S120, the path from the starting position to the target position may be searched for through locations according to the target search direction. The target search direction may be dynamically updated during the process of searching for through locations.

[0081] In related technologies, it is usually necessary to check all nodes and paths by traversing, and in some cases, back propagation is required to update the planned path, which results in a large amount of calculation. Taking the traditional Monte Carlo tree search as an example, the Monte Carlo tree search includes four steps:

[0082] (1) Selection: Based on the current node, select the node with the largest UCB value that has not been traversed to visit. Here, Where v is the current node; v′ is the child node of the current node; Q(·) is the total reward value of the node; N(·) is the number of times the node is visited; C is a constant, for example ).

[0083] (2) Expansion: Perform random operations on the node selected in step (1) and develop a new child node in the grid graph.

[0084] (3) Simulation: Take the new node created in step (2) as the starting point, repeat steps (1) and (2) to select new leaf nodes to expand the search tree until the target point is reached or the specified exploration depth is reached.

[0085] (4) Back-propagation: When the search tree is expanded to the target point, it traverses backward from the target point to the root node, transmits the simulation result of step (3) to the root node, and updates the reward value, number of visits or other information of each node on the back-propagation path.

[0086] The mechanism of Monte Carlo tree search is to start from the starting position and use a greedy algorithm to select the best child node as the leaf node of the decision number to maximize the cumulative reward. In such a search process, a large number of irrelevant nodes that deviate greatly from the optimal path will be included, resulting in a decrease in optimization efficiency. As the scale of the target environment increases, this disadvantage becomes more and more obvious, and even leads to a "dimensionality disaster", making it impossible to achieve timely path planning and making it impossible to use automatic transport devices to perform transportation.

[0087] In response to such a problem, the embodiments of the present disclosure recognize that irrelevant nodes that deviate greatly from the optimal path can be eliminated by limiting the search direction, so as to improve the optimization efficiency.

[0088] Specifically, in step S120, the via position may be searched in the following manner: starting from the starting position, searching for the via position that satisfies the target search direction. Here, satisfying the target search direction may mean that the direction from the reference position to the via position is the same as the target search direction or the same as the component of the target search direction.

[0089] by Figure 4 For example, during the first search, the starting position is used as the current position of the automatic transport device. Specifically, for position 2 as the starting position and position 18 as the target position, the target search direction may include: the direction from position 2 to position 18; and / or, from position 2 to the positive direction of the x-axis and the positive direction of the y-axis. Therefore, the search is started from position 2, and the adjacent positions 1, 3 and 7 are located in the negative direction of the x-axis, the positive direction of the x-axis and the positive direction of the y-axis of position 2, respectively. It can be seen that position 1 does not meet the target search direction, while position 3 and position 7 meet the target search direction. Therefore, position 3 and position 7 can be used as pass-through positions, and based on position 3 and position 7, the search for other pass-through positions can be continued according to the target search direction until the target position 18 is reached.

[0090] Taking position 3 as an example, the last searched passing position 3 can be used as the current reference position, and the current target search direction can include: the direction from position 3 to position 18 (i.e., from position 3 to the positive direction of the y-axis). Therefore, the search starts with position 3 as the starting point, and the adjacent positions 2, 4, and 8 are located in the negative direction of the x-axis, the positive direction of the x-axis, and the positive direction of the y-axis of position 3, respectively. It can be seen that positions 2 and 4 do not meet the current target search direction, and position 8 meets the current target search direction. Therefore, position 8 can be used as the currently searched passing position, and based on position 8, the target search direction for the next search (for example, the direction from position 8 to position 18) can be determined, and other passing positions continue to be searched until the target position 18 is reached.

[0091] As an example, starting from the starting position, the search for expanded subnodes (or passing positions) may be performed step by step for all optional positions in the target environment until the target position is reached.

[0092] For example, starting from the starting position, the step of searching for a passing position that satisfies the target search direction may include: performing a search operation on an optional position in the target environment until the target position is found. Here, the search operation may be performed in the following manner:

[0093] like Figure 5 As shown, in step S510, a first position that satisfies an adjacency condition with the current reference position may be determined from optional positions in the target environment.

[0094] Here, the reference position of the first search operation may be the starting position, and the adjacency condition may include: the first position is adjacent to the current reference position, and there is no obstacle between the first position and the current reference position, wherein the adjacency may refer to the direction of dividing the grid (for example Figure 4The absence of obstacles between two positions may mean that: there are no obstacles at both positions (such as operating stations, material storage areas, etc.); or both positions are accessible to automatic conveying devices. Whether there is an obstacle at each position may be determined based on the state transition probability matrix as described above. For example, δ=1 may indicate that there is an obstacle at the position; δ=0 may indicate that there is no obstacle at the position. Figure 4 Taking position 2 as the current reference position as an example, when there are no obstacles at position 2 and positions 1, 3 and 7, positions 1, 3 and 7 can be determined as the first position. As an example, the optional position can be a position in the target environment other than the current reference position, or can be a pre-specified position in the target environment.

[0095] In step S520, a second position satisfying a direction condition with respect to the current reference position may be determined from the first position.

[0096] Here, the direction condition may include: the direction from the current reference position to the second position is the same as the current target search direction or is the same as the component of the current target search direction.

[0097] by Figure 4 For example, when the current target search direction is the direction from position 2 to position 18, its components may be the positive direction of the x-axis and the positive direction of the y-axis. When the current reference position is position 2 and the corresponding first positions include position 1, position 3 and position 7, the direction from position 2 to position 1 does not satisfy the direction condition, while the directions from position 2 to position 3 and to position 7 satisfy the direction condition. Therefore, position 3 and position 7 may be determined as the second position.

[0098] In step S530, in response to the second position being different from the target position, the second position can be determined as a passing position, and the second position can be used as a reference position for the next search operation, and the next search operation can be performed.

[0099] In step S540, in response to the second position being the same as the target position, the search operation may be terminated.

[0100] Specifically, when the second position is not the target position, it can be considered that the search to reach the target position has not been completed yet, and the search needs to be continued until the second position is found to be the target position. The passing positions from the starting position to the target position can be determined based on the previously performed search process.

[0101] by Figure 4For example, the second positions include position 3 and position 7, neither of which is the target position, therefore, position 3 and position 7 can be used as reference positions for the next search, and the above steps S510 to S540 are performed again. Here, for multiple second positions, the search operation can be performed separately, and the order of each search operation can be recorded, so as to determine the entire path according to the order of the searched passing positions.

[0102] In the above process, starting from the starting position, the current target search direction can be determined according to the reference position of each search, and the search of each passing position can be gradually expanded to determine the path to the target position. This is very beneficial for improving the search speed. In particular, for transportation scenarios in large places, the efficiency of path planning can be improved. For example, when the target environment is the assembly environment of a wind turbine, since the wind turbine is a large equipment, the size of its components is large (for example, the length of its blades is usually tens of meters or even hundreds of meters). Therefore, the area occupied by its assembly environment is usually large. Therefore, it is very necessary to quickly realize the calculation of path planning through the above method.

[0103] An example implementation method for implementing the above-mentioned step of determining the passing position is given below.

[0104] Specifically, it can be assumed that the target position of the automatic transport device is (G x ,G y ), the current reference position is (V x ,V y ), the child node to be expanded (such as the second position above) is (V x ′,V y ′), the method of sub-node selection can be expressed by equations (3-1) to (4-2):

[0105] G x >V x And G y <V y ,but

[0106] G x >V x And G y >V y ,but

[0107] G x <V x And G y >V y ,but

[0108] G x <V xAnd G y <V y ,but

[0109] The above formulas (3-1) to (4-2) respectively indicate that: when the target position is located in the upper right / lower right / lower left / upper left direction of the current position of the automatic transport device, according to the target search direction, only the position located in the upper right / lower right / lower left / upper left direction of the current reference position is considered as the second position (or passing position).

[0110] Although examples of expressions for determining passing positions are given above, their implementation is not limited to the above examples. Expressions for determining passing positions can also be constructed based on the actual rules for establishing a coordinate system, as long as a second position corresponding to the current reference position can be determined to gradually determine the passing position.

[0111] In addition, as an example, in an embodiment of the present disclosure, the speed of searching for passing locations can be further improved by narrowing the search range.

[0112] Specifically, in step S120, before executing the above-mentioned process of searching for passing positions, it can also include: dividing a target area in the target environment according to the starting position, the target position and the environmental boundary of the target environment; searching for passing positions in the target area according to the target search direction.

[0113] Here, the target area may include a starting position and a target position, and the range of the target area may be smaller than the range of the target environment.

[0114] Specifically, in some cases, the scope of the target environment may be large, while the starting position and the target position may be close to each other. Therefore, when planning the path, there is no need to consider the entire target environment. Instead, the target environment can be divided into a local area including the starting position and the target position for planning. This can improve the planning speed and avoid unnecessary node searches.

[0115] As an example, the target area can be determined by:

[0116] like Figure 7 As shown, in step S710, a first candidate position among the starting position and the target position that is closest to the environment boundary in a preset direction may be determined.

[0117] As an example, the preset direction may be, for example, the direction of the boundary of the target environment, or the grid direction of the gridded target environment, or the axis direction of the coordinate system constructed in the target environment, etc. Here, there may be multiple preset directions, for example, four. Each preset direction may be used to determine at least one boundary of the target area.

[0118] by Figure 6 For example, the preset directions may include a first direction, a second direction, a third direction and a fourth direction, and the first direction, the second direction, the third direction and the fourth direction may be respectively: two relative directions in the length and two relative directions in the width of the target environment; or, two relative directions in the row direction and two relative directions in the column direction of the grid in the target environment; or, the positive and negative directions of the x-axis and the positive and negative directions of the y-axis of the coordinate system constructed in the target environment.

[0119] Assume that the starting position is position 27 and the target position is position 13. For such a starting position and target position, in the first direction (for example Figure 6 The first environment boundary closest to the target environment (e.g. Figure 6 The first candidate position of the left boundary of the 2nd direction (for example, the left boundary of the 2nd direction) is the starting position (i.e., position 27); Figure 6 The second environment boundary closest to the target environment (e.g. Figure 6 The first candidate position of the right boundary of the target direction (i.e., position 13) is the target position; in the third direction (e.g. Figure 6 The third environment boundary closest to the target environment (e.g. Figure 6 The first candidate position of the upper boundary of the fourth direction (eg Figure 6 The fourth environment boundary closest to the target environment (e.g. Figure 6 The first candidate position of the lower boundary of the image is the starting position (i.e., position 27).

[0120] In step S720, a second candidate position at a preset distance from the first candidate position may be determined along a preset direction.

[0121] Here, the preset distance can be set according to actual needs. For example, the preset distance can be a preset grid. For different preset directions, the preset distance can be the same or different.

[0122] by Figure 6 For example, the preset distance corresponding to the first direction may be 2 grids. Therefore, along the first direction, the second candidate position that is a preset distance away from the first candidate position (position 27) is determined to be position 25.

[0123] In step S730, based on the positional relationship between the second candidate position and the environment boundary, a boundary of the target area in a preset direction may be determined to determine the target area.

[0124] In this step, when the position at a preset distance from the first candidate position does not exceed the environmental boundary of the target environment, the row or column where the second candidate position is located can be used as the edge row or column of the target area in the corresponding preset direction; when the position at a preset distance from the first candidate position exceeds the environmental boundary of the target environment, the environmental boundary of the target environment can be used as the boundary of the target area in the preset direction.

[0125] by Figure 6 For example, the preset distances corresponding to the first direction to the fourth direction may be 2 grids respectively. Figure 6 In the first direction, the second candidate position is position 25, and the column where position 25 is located can be used as the edge column of the target area in the first direction. In the second direction, the second candidate position is position 15, and the column where position 13 is located can be used as the edge column of the target area in the second direction. In the third direction, the second candidate position exceeds the environmental boundary of the target environment (i.e., the upper boundary), and the environmental boundary of the target environment in the third direction can be used as the boundary of the target area in the third direction. Similarly, in the fourth direction, the second candidate position exceeds the environmental boundary of the target environment (i.e., the lower boundary), and the environmental boundary of the target environment in the fourth direction can be used as the boundary of the target area in the fourth direction.

[0126] Through the above method, on the one hand, the target area can be divided from the target environment to improve the path planning speed; on the other hand, in the above process, by introducing a preset distance, a certain distance can be expanded outward relative to the first candidate position, thereby avoiding the situation where there is no solution in the area after the target area is divided.

[0127] Specifically, dynamic programming has two characteristics: one is the optimal substructure, that is, the optimal solution to a problem can be composed of the optimal solutions to several small problems; the other is the repeated substructure, that is, by finding the recursive relationship between the states of the subproblems. One of the most effective ways to speed up sample access is to delete invalid samples and reduce the number of training samples. If only the starting position and target position of the automatic transport device are considered without considering the relative position to the boundary of the environment, the optimal path cannot be found in some cases.

[0128] like Figure 6 As shown in the figure, the starting position of the automatic transport device is 27 (x=3, y=3), the target position is 13 (x=5, y=1), and the dotted line represents the feasible path from the starting position to the target position. If the preset distance is not expanded outward based on the first candidate position, the target area obtained by the division is area 61. In this case, if there are obstacles at positions 19, 20, 21, 28, 29 and 30, a feasible path cannot be planned based on area 61.

[0129] In this regard, in the embodiments according to the present disclosure, since the target area can be obtained by expanding outward from the first candidate position to the second candidate position, a certain searchable area is left outside the starting position and the target position. For example, when the starting position and the target position are not located at the environmental boundary of the target environment, in the planned target area, the starting position and the target position will not be located at the boundary position of the target area, so that each adjacent position of the starting position and the target position can be divided into the target area, thereby avoiding the problem of not being able to search for a feasible path after dividing the target area. For example, Figure 6 In FIG. 6 , the target area obtained by the area segmentation method according to the embodiment of the present disclosure is area 62. Even if there are obstacles at positions 19, 20, 21, 28, 29, and 30, a feasible path from position 27 to position 13 can be planned based on area 62 (e.g., Figure 6 ).

[0130] In the embodiment of the present disclosure, according to the relative position of the starting position of the automatic transport position, the target position and the environment boundary, a specific target area is divided to exclude irrelevant nodes and speed up the expansion of leaf nodes. An example implementation of the above step of determining the target area will be described below.

[0131] Specifically, the starting position of the automatic transport position can be set as (S x ,S y ), the target position is (G x ,G y ), the horizontal coordinate of the left border of the target environment is X l , the horizontal coordinate of the right boundary is X r , the ordinate of the upper boundary is Y u , the ordinate of the lower boundary is Y d The location of the target area is represented as follows: is the upper left coordinate, is the upper right coordinate, is the lower left coordinate, is the lower right coordinate, there are OK The preset distances in each preset direction may be d (for example, 2), so the coordinates of the four corner positions of the target area may be obtained as follows:

[0132]

[0133]

[0134] Although the above describes an example method of dividing the target area, the embodiments of the present disclosure are not limited thereto, and the division may be performed in other ways, for example, the target area may be divided by extending a predetermined distance outward according to the starting position and the target position. In addition, although the above formulas show by way of example that the preset distances in each preset direction may all be d, the embodiments of the present disclosure are not limited thereto, and the preset distances in at least a portion of the preset directions may also be different.

[0135] Return to reference Figure 1 In step S130, a possible path from the starting location to the target location can be planned based on the passing locations.

[0136] By gradually expanding and determining each passing position until the target position in the above manner, a feasible possible path can be planned. In some cases, there may be one or more possible paths, and in subsequent steps, the final path can be determined from these possible paths; in some cases, it may be impossible to plan a possible path, in which case, the current path planning can be directly exited.

[0137] In the traditional Monte Carlo tree search method, since the search direction is not limited, all nodes need to be considered when searching for a feasible path, resulting in a large amount of calculation and the need for back propagation to determine a feasible path. Figure 8 The traditional Monte Carlo tree is shown for Figure 4 The result diagram of the target environment search from position 2 to position 18 is as follows: Fig. 9 The path planning method according to the embodiment of the present disclosure is shown. Figure 4 The result diagram of the target environment search for the passing position is shown in Fig. 9 In the figure, the dotted lines represent leaf nodes existing in the traditional Monte Carlo tree, but in the path planning method according to the embodiment of the present disclosure, these nodes will not be searched as passing positions.

[0138] By comparison Figure 8 and Fig. 9 It can be seen that in the traditional search method, there are many redundant leaf nodes, and these leaf nodes cannot reach the target position; while in the method disclosed in the present invention, 4 reachable paths can be directly obtained: (1) 2→3→8→13→18; (2) 2→7→8→13→18; (3) 2→7→12→13→18; (4) 2→7→12→17→18. There is no need for back propagation. All reachable paths can be determined through unidirectional expansion, and all available nodes can be quickly located to avoid introducing irrelevant nodes. The calculation process is simple and fast.

[0139] Here, in the final assembly environment of a wind turbine, due to the large area and large number of nodes, it would take a lot of time to backtrack to find the shortest path after finding all feasible paths. In the embodiment of the present disclosure, by adopting a single-step update search, there is no need for a back propagation process, which saves time. This has obvious advantages in the final assembly environment of a wind turbine and can significantly reduce the calculation time.

[0140] In step S140, a target path for the automatic transport device from the starting position to the target position may be determined from the possible paths.

[0141] In this step, when there is only one possible path, the possible path can be used as the target path.

[0142] In the case where there are multiple possible paths, the optimal path among the possible paths may be used as the target path. For example, step S140 may include the following steps:

[0143] like Fig.10 As shown, in step S1010, the path coefficient of each possible path can be determined.

[0144] Here, the path coefficient can represent: the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced.

[0145] As an example, the distance traveled by the automated transport device on a possible path may be determined by summing up the distances of each straight-line driving segment, for example, it may be represented by the total number of grids passed through.

[0146] In an embodiment of the present disclosure, considering that a turn may cause a decrease in driving speed, the number of turns may be added to the advantage evaluation of a possible path.

[0147] Specifically, since there may be more than one path with the shortest total driving distance from the starting position to the target position of the automatic transport device, and it can only turn at nodes, and turning requires deceleration / stopping / rotation / acceleration, the fewer turns the automatic transport device makes, the shorter the path time, and the turning time increases with the increase of the turning angle.

[0148] As an example, it can be determined that the automatic conveyor experiences a turn by: in response to an angular difference between the direction of travel of the automatic conveyor when it moves to the current position and the direction of travel of the automatic conveyor when it moves to the next position, determining that the automatic conveyor experiences a turn when it moves from the current position to the next position.

[0149] Specifically, if there is an angle difference, it can be determined that the movement from the current position to the next position involves a turn; if there is no angle difference, it can be determined that the driving direction (or forward direction) from the current position to the next position does not change, and does not involve a turn. In this way, the turning position of the automatic conveyor can be quickly determined based on whether the angle of the driving direction changes.

[0150] As an example, the shortest path can be found by the following method: (1) before the automatic transport device sets off, the angle at which it enters the environment is compared with the direction of the alternative paths, and the path with the smallest absolute angle deviation is found as the starting path; (2) when there is more than one alternative path in front of the automatic transport device, the angle deviation between the direction of the automatic transport device reaching the node and the driving direction of the next segment of the path is calculated, and the path with the smallest absolute angle deviation is selected in real time as the next segment of the driving path.

[0151] by Fig.11 For example, the starting position is position 0 and the target position is position 29. Fig.11 As shown in the left figure, the arrow at the entrance indicates the initial direction of the automatic conveyor. Fig.11 In the example on the left, there are two optional paths (indicated by dashed arrows and solid arrows respectively). The starting path is determined by comparing the initial direction of the automatic conveyor with the angle of each path: in, It represents the angle of path 1 with node 0 as the reference. It represents the angle of path 2 with node 0 as the reference; Indicates the current angle of the automatic conveyor at node 0. The absolute value of the angle deviation between the current direction of the automatic conveyor and path 1 is 0°, while the absolute value of the angle deviation between the current direction of the automatic conveyor and path 2 is 90°.

[0152] Similarly, for node 2 (x=2, y=0), if Fig.11 As shown in the right figure, the current direction of the automatic transport device can be from node 1 to node 2, and there are two shortest paths. The current angle of the automatic transport device can be compared with the angle of each path in real time to determine whether there is a turn. The angle of the automatic transport device starting from node 2 (x=2, y=0) is: in, represents the angle of path 3, represents the angle of path 4; Represents the angle of the AGV at node 2. The absolute value of the angle deviation between the current direction of the automatic transport device and path 3 is 0°, while the absolute value of the angle deviation between the current direction of the automatic transport device and path 4 is 90°.

[0153] The above describes the process of determining the distance traveled by the automatic transport device and the number of turns it has experienced. In this way, the path coefficient can be determined based on the distance and the number of turns. For example, a unit distance can be set, and the path coefficient can increase by a first preset value (for example, 1) for each additional unit distance; and the path coefficient can increase by a second preset value (for example, 1) for each additional turn. Here, the first preset value and the second preset value can be the same or different. In this way, the final path coefficient can be determined.

[0154] It should be noted that the above reference Fig.11 Describes the process of determining in real time whether there is a turn at each waypoint during the search, such as in the process of determining possible paths (such as in determining Fig. 9 In the process of expanding the relationship), each time a passing position is expanded, it is determined whether there is a turn when reaching the passing position; it is also possible to determine whether there is a turn when reaching each passing position for each possible path after determining the possible paths.

[0155] Return to reference Fig.10 In step S1020, the possible path with the smallest path coefficient can be determined as the target path.

[0156] When the path coefficient is determined, the path with the smallest path coefficient among all possible paths can be used as the final target path.

[0157] by Fig. 9 For example, Move can be used to represent the path coefficient, and Turn can be used to represent the turn. Fig. 9 The Move values ​​of the four reachable paths are: (1) 2→3→8→13→18 (Move=5); (2) 2→7→8→13→18 (Move=6); (3) 2→7→12→13→18 (Move=6); (4) 2→7→12→17→18 (Move=6). Paths (2) and (3) can be deleted according to the principle of minimum Move value, and the optimal paths are paths (1) and (4).

[0158] In this way, the optimal path can be more accurately selected as the target path by considering both the distance and the number of turns. It should be noted that the path coefficient can be calculated for each possible path after the possible path is determined, or it can be calculated in the process of determining the possible path (for example, in determining the path coefficient). Fig. 9 In the process of expanding the relationship, the path coefficient is updated every time a passing position is expanded, so that the path coefficient of the possible path is obtained when the target position is expanded.

[0159] Here, the target path may be one or more. In the case where there are multiple target paths, any one of the paths may be used as a planned path to control the automatic transport device to travel along the path.

[0160] In the above step S1010, it is described that the path coefficient of the possible path is determined based on the distance and number of turns experienced by the automatic transport device and the first preset value and the second preset value. In another example, for the final assembly environment of large equipment, the number of turns can be weighted to improve the stability of equipment transportation.

[0161] Specifically, taking the target environment as the final assembly environment of a wind turbine generator set as an example, due to the larger size of the components of the wind turbine generator set, compared with other application environments, the size of the automatic transport device and the transported materials is larger, and the distance between adjacent nodes is also farther. Therefore, the automatic transport device takes more time and energy to travel at multiple nodes. In the final assembly environment, the automatic transport device makes turns more complicated, and needs to be more cautious and more time-consuming and energy-consuming during the turning process. Therefore, the weight of the number of turns in the path of the automatic transport device is greater than in other environments (such as logistics scenarios). In the final assembly environment of a wind turbine generator set, the automatic transport device needs to reduce turns as much as possible to achieve more stable transportation.

[0162] As an example, in step S1010, the step of determining the path coefficient of each possible path may include: determining the distance coefficient and the turning coefficient respectively according to the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; determining the turning weight according to the area size of the final assembly environment; using the turning weight, weighting the turning coefficient to obtain a weighted turning coefficient; and obtaining the path coefficient based on the distance coefficient and the weighted turning coefficient.

[0163] Here, the distance coefficient may be determined, for example, according to the preset unit distance and the distance traveled as described in the above example, and the distance coefficient may increase by a first preset value (e.g., 1) for each additional unit distance, and as an example, the distance coefficient may be the number of grids. For each additional turn, the turn coefficient may increase by a second preset value (e.g., 1), and here, the first preset value and the second preset value may be the same or different.

[0164] The turning weight may be positively correlated with the area size of the final assembly environment. When the area size is larger, the turning weight may be set larger to minimize the number of turns in the final target path and maintain the stability of the transportation. For example, a preset conversion relationship may be set to determine the corresponding turning weight according to the area size of the current environment to weight the turning coefficient. In this way, the sum of the distance coefficient and the weighted turning coefficient may be used as the path coefficient.

[0165] An example process of determining a target path of an automatic transport device according to an embodiment of the present disclosure is described above, and an example implementation manner of implementing the process will be given below.

[0166] Specifically, for Figure 5 The search operation shown can be implemented by the following Algorithms 1 and 2 to determine whether there is an obstacle in the current forward direction.

[0167] Specifically, if there are no obstacles in the forward direction of the automatic transport device, a method of establishing a regional adjacency list can be used, as shown in Algorithm 1, where "G'[·][·]:G'[*][*]→adj[]" means: if node G'[·][·] is adjacent to node G'[*][*], node G'[·][·] can be added to adj[].

[0168] In Algorithm 1, the grid-arranged nodes (or positions) in the target environment can be converted into an obstacle-free adjacency list. The pseudo code of Algorithm 1 can be shown in Table 1 below:

[0169] Table 1

[0170]

[0171] Similarly, if there are obstacles in the forward direction of the automatic conveying device, a regional adjacency list can be established, as shown in Algorithm 2, where obstacle list[] represents a list of obstacle locations; “G’[·][·]: G’[*][*]→adj_obs[]” means that node G’[·][·] is adjacent to node G’[*][*], and node G’[·][·] can be added to the adjacency list adj_obs[].

[0172] In Algorithm 2, the grid-arranged nodes (or positions) in the target environment can be converted into an obstructed adjacency list. The pseudo code of Algorithm 2 can be shown in Table 2 below:

[0173] Table 2

[0174]

[0175]

[0176] For the search operation, we can find the path with the shortest distance and the fewest turns from the starting position to the target position: on the one hand, we can use heuristic search instead of blind search, limit the expansion range of the search tree according to the positions of the starting position and the target position, reduce the number of irrelevant nodes, and improve the efficiency of the algorithm; on the other hand, we can advance the node evaluation process, increase the penalty factor for the turns of the automatic conveyor, and score the process each time a node is expanded to avoid wasting time in the back propagation process.

[0177] As an example, an example of pseudo code for searching for the passing positions and the final target path is shown in Table 3 below:

[0178] Table 3

[0179]

[0180]

[0181] Although the pseudo code examples for determining the passing position and performing the search operation are given above, the implementation thereof is not limited to the above example, and may also be implemented according to other code logics.

[0182] The above describes the path planning process of a single automatic transport device. According to the embodiments of the present disclosure, path planning can also be performed for multiple automatic transport devices in the same target environment.

[0183] Specifically, there may be multiple automatic transport devices, each of which corresponds to its own target position. Here, the target positions corresponding to different automatic transport devices may be different.

[0184] The path planning method according to an embodiment of the present disclosure may further include:

[0185] like Fig.12 As shown, in step S1210, the initial path of each automatic transport device may be determined based on the number of target paths corresponding to each automatic transport device;

[0186] In response to the existence of conflicting path segments between the initial paths of the automatic conveying devices, in step S1220, the initial paths of the automatic conveying devices with conflicts may be updated by performing a path update operation until there is no conflict between the initial paths of the automatic conveying devices, thereby obtaining final paths of the automatic conveying devices.

[0187] In response to the fact that there are no conflicting path segments between the initial paths of the automatic conveying devices, in step S1230, the current initial path of each automatic conveying device may be used as the final path of each automatic conveying device.

[0188] Here, the existence of conflict between the target paths of the automatic transport devices means that the automatic transport devices will be in the same path segment at the same time when traveling along their respective target paths; the absence of conflict between the target paths of the automatic transport devices means that the automatic transport devices will not be in the same path segment at the same time when traveling along their respective target paths.

[0189] Specifically, the conflict path segment refers to the part of the path that contains the node where the conflict occurs and is repeated. It can include multiple locations. The center position of the path conflict segment can be the exact location where the two automatic transport devices collide. In reality, the automatic transport devices cannot be regarded as nodes, and a safe distance must be maintained between the automatic transport devices. Therefore, it is only necessary to determine the conflict path segment where the conflict occurs and ensure that no two automatic transport devices appear at the same time in this segment.

[0190] As an example, the initial path of each automatic transport device can be determined in the following way:

[0191] In response to the number of target paths of the first automatic conveyor being less than the number of target paths of the second automatic conveyor, taking a target path among the target paths of the second automatic conveyor that does not overlap with an initial path of the first automatic conveyor as an initial path of the second automatic conveyor;

[0192] In response to the number of target paths of the first automatic conveying device being equal to the number of target paths of the second automatic conveying device, initial paths of the first automatic conveying device and the second automatic conveying device are determined from their respective target paths, so that there is no overlap between the initial paths of the first automatic conveying device and the second automatic conveying device.

[0193] Specifically, the path planning method for a single automatic conveyor described above can be used to obtain the shortest distance-time path (i.e., the above-mentioned target path) of each automatic conveyor. Here, the initial path of each automatic conveyor can be determined according to the number of target paths. For example, the initial path of each automatic conveyor can be determined in the order of the number of target paths from small to large. Specifically, the initial path of the automatic conveyor with the least number of target paths can be determined first, and the initial paths of other automatic conveyors can be determined in the order of the number of target paths from small to large. In the process of determining the initial path of each automatic conveyor, in response to the existence of a target path in the target path of the current automatic conveyor that does not conflict with the initial path of the determined automatic conveyor, the non-conflicting target path is used as the initial path of the current automatic conveyor; in response to the absence of a target path in the target path of the current automatic conveyor that does not conflict with the initial path of the determined automatic conveyor, any target path in the target path of the current automatic conveyor can be used as the initial path, or the target path in the target path of the current automatic conveyor with the shortest path segment that conflicts with the determined initial path can be used as the initial path.

[0194] For example, for any two automatic conveying devices, when the number of target paths of the first automatic conveying device is less than the number of target paths of the second automatic conveying device, the initial path of the first automatic conveying device can be determined first, and the target path that does not overlap with the initial path of the first automatic conveying device can be searched in the target path of the second automatic conveying device as the initial path of the second automatic conveying device. When the number of target paths of the first automatic conveying device is the same as that of the second automatic conveying device, the target paths of the two can be compared, and the path that does not overlap can be used as the initial path of the two. In this way, the initial path of each automatic conveying device can be determined in order from the highest priority (the smallest number of target paths) to the lowest priority (the largest number of target paths).

[0195] By adopting the above method, it is possible to avoid conflicting sections between the initial paths of the first automatic conveying device and the second automatic conveying device as much as possible.

[0196] For example, for the first automatic conveying device AGV1 and the second automatic conveying device AGV2, the initial shortest path of the automatic conveying device can be determined according to the number of target paths of the two. The specific method can be shown in Table 4 below:

[0197] Table 4

[0198]

[0199]

[0200] Here, although the above reference Table 4 shows the selection of initial paths for two automatic conveying devices, in an embodiment of the present disclosure, when there are more than two automatic conveying devices, the initial paths can be determined in pairs using the method of Table 4; or, the initial path of the current automatic conveying device can be determined in the order of the number of target paths from small to large using the method of Table 4, with reference to the initial path of the previous automatic conveying device in sequence. For example, the initial path of the automatic conveying device in the first position can be sorted first, and then the initial path of the automatic conveying device in the first position can be referred to, and the initial path of the second automatic conveying device can be determined according to the rules shown in Table 4, and so on.

[0201] In addition, in some cases, it may be impossible to avoid path conflicts between the two automatic conveying devices. For example, when the initial path of the first automatic conveying device is determined first, all target paths of the second automatic conveying device overlap with the initial path of the first automatic conveying device. In this case, the target path of the second automatic conveying device that has the least overlap with the initial path of the first automatic conveying device (or has the shortest conflicting path segment) can be used as the initial path. Alternatively, if the first and second automatic conveying devices have the same number of target paths and each has multiple target paths, and there are conflicting path segments between their target paths, the two target paths with the shortest conflicting path segments between the first and second automatic conveying devices can be used as their respective initial paths.

[0202] Return to reference Fig.12 , by detecting whether there is a conflict between the initial paths of each automatic conveying device, the initial path of each automatic conveying device can be updated, and based on the updated initial path, the next path update operation is performed until the initial paths of all automatic conveying devices do not conflict, and then the final path of each automatic conveying device is output.

[0203] Specifically, whether there is a conflict between the initial paths of each automatic conveying device can be detected in the following manner: determine the path position passed by the initial path of each automatic conveying device; and detect the automatic conveying devices whose initial paths have conflicting path segments by comparing the path positions passed by the initial paths of every two automatic conveying devices.

[0204] Here, the path position may refer to all positions on the initial path including the starting position and the target position. The overlapping path segments between the initial paths of each two automatic conveying devices, i.e., conflicting path segments, may be detected. The existence of conflicting path segments between the initial paths of the automatic conveying devices may mean that there is a common path segment between the initial paths of multiple (e.g., two) automatic conveying devices, and the multiple automatic conveying devices will reach the common path segment at the same time.

[0205] After the initial path is determined according to the quantity of the target path, there may still be conflicts between the initial paths. In this way, the initial paths can be compared and the conflicting paths can be updated until there is no conflict between the target paths of the automatic transport devices. By means of a cyclic solution, path planning between multiple automatic transport devices can be achieved.

[0206] As an example, the initial paths of the conflicting automatic transport devices may be updated in the following manner: the initial path of at least one target transport device may be updated by adjusting the time at which at least one of the conflicting target transport devices reaches the conflicting path segment.

[0207] In the above process, when there is a conflict in the initial path, the initial path of the target transport device can be updated in the following way:

[0208] like Fig.13 As shown, in step S1310, the distance index between the starting position and the target position of each of the conflicting target transport devices may be determined.

[0209] Here, the distance indicator may be used to characterize the distance between the starting position and the target position, and may be, for example, but not limited to, Manhattan distance or other distance metrics.

[0210] In step S1320, for any two conflicting target transport devices, the initial paths of the two conflicting target transport devices may be determined by performing the following update operations:

[0211] In response to the target position of the first target transport device being located on the initial path of the second target transport device, based on the length of the conflicting path segment between the first target transport device and the second target transport device, the moving speed of the first target transport device, and the preset dwell time, determining a waiting time for the second target transport device before entering the conflicting path segment, so as to update the initial path of the second target transport device;

[0212] In response to the target position of the first target conveying device not being located on the initial path of the second target conveying device, based on the length of the conflicting path segment between the first target conveying device and the second target conveying device and the moving speed of the automatic conveying device, the waiting time of the second target conveying device when entering the conflicting path segment is determined to update the initial path of the second target conveying device.

[0213] Here, the distance index of the first target transport device may be greater than the distance index of the second target transport device.

[0214] Specifically, when there is a conflict in the initial path, in order to ensure that the total system time for all target transportation devices to complete a task is minimized, priority can be given to modifying the path or schedule of the target transportation device with a smaller distance index to shorten the overall transportation time of all target transportation devices as much as possible.

[0215] Taking Manhattan distance as an example, when the target position of the first target transportation device with a large Manhattan distance is on the shortest path (or initial path) of the second target transportation device with a small Manhattan distance, the second target transportation device can be made to stop and wait for a period of time before entering the conflicting path segment, so that the first target transportation device can pass through the conflicting path segment first. In this way, on the one hand, path planning conflicts can be avoided, and on the other hand, the overall transportation time can be minimized.

[0216] Here, the waiting time of the second target conveying device before entering the conflicting path segment may be determined based on the length of the conflicting path segment between the first target conveying device and the second target conveying device, the moving speed of the first target conveying device, and the preset dwell time.

[0217] As an example, the moving speeds of the automatic transport devices may be the same or different, and the dwell time may refer to the time required for the first target transport device to perform an operation at the target location, such as loading or unloading materials, etc. The dwell time may be predetermined according to the requirements of the transport task. In this way, the time required for the first target transport device to pass through the conflicting path segment and the dwell time required at the target location may be determined, so that the sum of the two may be used as the waiting time for the second target transport device before entering the conflicting path segment.

[0218] For example, the waiting time T of the second target transport device before entering the conflicting path segment can be expressed by the following formula (5):

[0219] T = Len_sec × t c +t w (s) (5)

[0220] Among them, Len represents the number of grids in the conflicting path segment, which can represent the length of the conflicting path segment. c represents the moving time of the first object transport device between adjacent grids, which can be determined based on the moving speed of the first object transport device and the length of each grid, for example, Among them, L n Indicates the length of each grid, L c represents the length of the automatic transport device (such as the first target transport device), and v represents the moving speed of the automatic transport device (such as the first target transport device). w (s) indicates the preset dwell time.

[0221] In an embodiment of the present disclosure, path planning may include the path of the automatic transport device and the waiting time during the driving process. In the above manner, the planning of the target path of the second target transport device can be updated. Although there is a conflicting path segment on the driving path between the first target transport device and the second target transport device, by adding a waiting time to the driving path of the second target transport device, the two transport devices can avoid conflict when passing through the conflicting path segment at the same time.

[0222] Still taking Manhattan distance as an example of the distance indicator, when the target position of the first target transportation device with a large Manhattan distance is not on the shortest path (or initial path) of the second target transportation device with a small Manhattan distance, the waiting time of the second target transportation device before entering the conflicting path segment can be determined based on the length of the conflicting path segment between the first target transportation device and the second target transportation device and the moving speed of the first target transportation device.

[0223] For example, the waiting time T of the second target transport device before entering the conflicting path segment can be expressed by the following formula (6):

[0224] T = Len_sec × t c (6)

[0225] Wherein, Len represents the number of grids of the conflicting path segment, which can represent the length of the conflicting path segment; t c represents the moving time of the first object transport device between adjacent grids, which can be determined based on the moving speed of the first object transport device and the length of each grid, for example, Among them, L n Indicates the length of each grid, L c Indicates the length of the automatic transport device (eg, the first target transport device).

[0226] In the above formula (6), since the target position of the first target transport device is not on the initial path of the second target transport device, the first target transport device will not stay in the conflicting path segment. Therefore, compared with the above formula (5), it is not necessary to consider the stay time of the first target transport device.

[0227] By means of the above method, the timetable of the final path of the second target transport device with a smaller distance index can be modified to obtain the optimal collision-free path planning scheme for all automatic transport devices.

[0228] Table 5 below shows an example of updating the target path based on the above method.

[0229] Table 5

[0230]

[0231] The updating operation in step S1320 may be used to update the initial paths of any two target transport devices that have conflicts, and the updating operation in step S1320 may be iteratively performed until there are no conflicts between the initial paths of all target transport devices.

[0232] In addition, for the path planning method of the above-mentioned multiple automatic transport devices, its advantages are particularly prominent in the example of the target environment being the final assembly environment of a wind turbine generator set. Specifically, due to the large range of the final assembly environment, the automatic transport device takes a longer time to travel and the size of the automatic transport device is also larger than that of other environments. According to the planning strategy proposed above, when two automatic transport devices cannot avoid path conflicts, one of the automatic transport devices can be made to wait before the collision until the other automatic transport device passes through the path-sensitive area (the above-mentioned conflicting path segment). While ensuring safe passage, the path-sensitive area is determined to determine the conflicting area that the two automatic transport devices need to avoid. Here, for the final assembly environment, the path segment is closer to an area (here referred to as the path-sensitive area). Through the above-mentioned planning method, the safety of the automatic transport device when passing through each area can be ensured to a higher degree, and collisions between large-sized automatic transport devices can be avoided.

[0233] In addition, in the final assembly environment, the automatic transport device takes a longer time to travel from the specified target location to the specified target location, and the total completion time of the system is determined by the longest path driving time. Therefore, it is necessary to make the automatic transport device with a small distance indicator or a short distance wait, so as to ensure that the automatic transport device with a long path can complete the task as soon as possible.

[0234] Tables 6 and 7 below show examples of algorithms for implementing collision-free path planning for the above-mentioned multiple automatic transport devices.

[0235] Specifically, the path conflict detection algorithm can be divided into two parts: a conflict pre-detection algorithm and an algorithm for determining conflict path segments (or conflict key segments). The pseudo code of the path conflict pre-detection algorithm is shown in Algorithm 4 in Table 6 below:

[0236] Table 6

[0237]

[0238] The above algorithm 4 can be used to determine whether there are common nodes (e.g., conflicting path segments) between the initial distance-time shortest paths of multiple automatic transport devices (i.e., the initial paths described above). If two paths contain the same nodes, the time when the relevant automatic transport devices arrive at these nodes is further compared according to the algorithm 5 shown in Table 7 below to determine whether there is a conflict between the two paths, thereby calculating the conflicting path segments. The pseudo code for determining the conflicting path segments is shown in Table 7 below:

[0239] Table 7

[0240]

[0241]

[0242] In Table 7, rows 1-3 indicate the time to reach the road section calculated based on the time it takes for the automatic transport device to reach the node in the path; rows 4-8 indicate the time to reach the node in the path of the automatic transport device. Comparisons are performed, where n represents the total number of automatic transport devices. In this way, the time list of each automatic transport device arriving at the nodes and sections can be determined, so that the nodes or sections corresponding to the repeated elements can be added to the list section[] and the length of the conflicting path segment can be obtained.

[0243] In the related art, the path conflict problem of multiple AGVs can also be avoided through the traditional time window method. However, compared with the traditional time window method, the method according to the embodiment of the present invention can, on the one hand, reduce the algorithm calculation complexity. Specifically, by calculating the dangerous area where the path conflict occurs, it can ensure that when one automatic transport device enters this area, other automatic transport devices can no longer enter, thereby avoiding the occurrence of collisions and leaving a safe distance for the movement of the automatic transport device.

[0244] On the other hand, the method improves the speed of conflict detection and the efficiency of the algorithm. Specifically, the method first compares the coordinates of the nodes in the initial path of the automatic transport device two by two. If there are identical nodes, it continues to calculate the length of the key segment, otherwise it exits, which significantly reduces the number of path comparisons. In the traditional time window method, the coordinates of every two nodes and the time when the automatic transport device reaches these nodes are compared. Although this improves the detection accuracy, it is not worth the loss in scenarios where high speed and stability requirements are required.

[0245] On the other hand, the method has simple classification and high practicality. Specifically, the method omits unnecessary subdivision of conflict types, reduces storage costs, improves algorithm robustness, and at the same time reduces the difficulty of algorithm design and increases practicality.

[0246] Fig.14A and Fig. 14B The complete process of an application example of the path planning method according to an embodiment of the present disclosure is shown, taking a multi-AGV scenario as an example.

[0247] like Fig.14A As shown, in step S1401, the starting point / target point of multiple AGVs can be determined. In step S1402, the target environment can be segmented according to the relative position of the starting point / target point and the boundary. For example, the starting point and target point of each AGV can be segmented to obtain their respective target areas. In step S1403, the nodes in the environment can be numbered, for example, 1, 2, 3...n, where the nodes can correspond to a grid.

[0248] In step S1404, it can be determined for each AGV whether there is an obstacle. In the case of an obstacle, in step S1405, the obstacle node position can be stored and an obstacle node position list (obstacle list) can be generated. In step S1406, the obstacle nodes can be excluded and an adjacency list of the segmented area that does not contain the obstacle point can be established. In step S1407, an adjacency list of an obstacle-free environment can be obtained. In step S1408, a search tree from the starting point to the target point can be established according to the target search direction. In this way, in step S1409, all feasible paths from the starting point to the target point can be obtained for each AGV.

[0249] In the absence of obstacles, in step S1410, an adjacency list of the segmented area can be directly established. In step S1411, a search tree from the starting point to the target point can be established according to the target search direction. In step S1412, the current position V can be determined. x-y and the target point G x-y For example, to determine whether the position relationship G is satisfied. x-y >=V x-y If the relationship is satisfied, the child node can be expanded according to the position relationship. For example, in step S1413, the expanded child node and the previous node satisfy: NODE+1 x-y >=NODE x-y ; In step S1414, the expanded child node and the previous node satisfy: NODE+1 x-y <NODE x-y Finally, in step S1409, all feasible paths from the starting point to the target point can be obtained for each AGV.

[0250] In step S1415, the scores (eg, path coefficients) of all feasible paths may be calculated, and the path with the least score may be obtained. Here, the rule for calculating the score may be determined according to the reward and penalty rules described above.

[0251] In step S1416, all nodes in the path can be traversed to initialize the time for the AGV to arrive at each node, and in step S1417, the path timetable can be updated according to the position of the turns in the path, and the time for the AGV to arrive at each position is recorded in the path timetable.

[0252] In step S1418, single AGV path planning can be completed to obtain the initial target path of each AGV, and based on these target paths, multi-AGV path planning can be started.

[0253] like Fig. 14B As shown, in step S1419, the initial path can be determined according to the number of target paths. For example, the initial path of the automatic transport device can be determined according to Table 4.

[0254] In step S1420, conflict pre-detection can be performed. Specifically, the shortest paths of the AGVs can be compared with each other (i.e., the target paths planned for a single AGV as described above). In step S1421, it can be determined whether each two AGVs (e.g., AGV i+1 With AGV i ) whether there are the same nodes, that is, to determine the conflicting path segment between the two AGVs. In the case that the two AGVs do not have the same nodes, step S1422 can be executed, i=i+1, and the above-mentioned conflict detection operation can be performed on the next pair of AGVs.

[0255] In the case of conflicting path segments, in step S1423, the Manhattan distance from the AGV starting point to the target point can be calculated. In step S1424, the conflicting path segments and lengths of the two AGVs can be calculated. In step S1425, it can be determined whether the target point of the AGV with a larger Manhattan distance is located on the path of the AGV with a smaller distance.

[0256] In the case of being on the AGV path with a small distance, in step S1426, the waiting time Len_sec×t before the AGV with a small distance in Manhattan departs c +t w (s); if it is not on the AGV path with a small distance, in step S1427, the waiting time before the departure of the AGV with a small Manhattan distance is Len_sec×t c (s).

[0257] In step S1428, the timetable for AGVs to pass through each node can be modified, and in step S1422, i=i+1 is set, and in step S1429, it is determined whether i=n, ​​where n is the total number of AGVs. In response to i≠n, the process can return to step S1420 to perform the above-mentioned conflict detection operation on the next pair of AGVs; in response to i=n, ​​in step S1430, the traversal of all AGVs is completed to obtain the multi-AGV conflict-free shortest path.

[0258] In addition, in order to increase the dynamics of the system, this method also proposes an automatic transport device scheduling method based on distance indicators. Specifically, by calculating the distance indicators to the locations of all alternative automatic transport devices, the automatic transport device that can reach the starting position of the task the fastest can be selected to perform the task, thereby shortening the idle time of multiple automatic transport devices.

[0259] Specifically, when the multi-AGV system receives a new task at time t, let the task starting point be (x0, y0). If there are p candidate AGVs (i.e., they have completed the transportation task and are in an idle state with a battery level higher than 20%), their locations are Loc = {(x1, y1), (x2, y2), …, (xp ,y p )}. The system calculates the Manhattan distance from the task starting point (x0, y0) in real time. The optimal AGV is the AGV corresponding to the minimum Manhattan distance, that is, argmin{D M}.

[0260] According to the path planning method of the embodiment of the present disclosure, a heuristic search method is proposed, which limits the expansion direction of the search tree by comparing the positional relationship between the current node and the target position, thereby accelerating the path optimization speed. In addition, in view of the problem that the traditional algorithm expands to the target position and then performs back propagation, which leads to low efficiency, this method advances the back propagation process and uses a single-step update to establish the search tree. In addition, a path evaluation standard based on the path coefficient is proposed, which scores all feasible paths from the starting position to the target position to find the shortest path with the least turns.

[0261] In addition, according to the path planning method of the embodiment of the present disclosure, the idea of ​​dynamic programming multi-stage optimization problem is introduced into the above-mentioned improved Monte Carlo tree search, the environment is divided into regions according to the starting position and the target position of the automatic transport device, and the path planning efficiency is increased by excluding nodes that are far away from the optimal path.

[0262] Through the above two improvements, the time complexity of the algorithm can be reduced to O(nlogn) level.

[0263] In addition, according to the path planning method of the embodiment of the present disclosure, an improved MDP modeling method can also be proposed for path planning problems such as automatic transport devices in wind turbine assembly environments, providing a modeling basis for the next step of automatic transport device path planning.

[0264] In addition, according to the path planning method of the embodiment of the present disclosure, an adaptive multi-automatic conveyor conflict-free path planning method based on the key segment of path conflict is also proposed. Specifically, after obtaining the distance-time shortest path of all AGVs, multiple distance-time shortest paths are pre-detected for conflicts, and then an adaptive method is used to obtain the final path and schedule. In addition, in order to increase the dynamics of the system, an automatic conveyor scheduling method based on distance indicators is proposed: by calculating the Manhattan distance to the locations of all alternative automatic conveyors, the automatic conveyor that can reach the starting position of the task the fastest is selected to perform the task, thereby shortening the idle time of multiple automatic conveyors.

[0265] The path planning method for a single automatic transport device and multiple automatic transport devices is described above. According to an embodiment of the present disclosure, after executing the path planning method, the planned path can also be evaluated to determine the planning quality of the target path.

[0266] In one example, the target path may be evaluated according to the total time taken on the target path.

[0267] Specifically, when the automatic transport device reaches any node, it can upload the current node coordinates to the server through wireless transmission methods such as Bluetooth and Wi-Fi, and the central control room can know the location of the automatic transport device in the environment. Taking the assembly environment of wind turbines as an example, the workstations in the assembly environment can be regarded as nodes, and the distance between each two adjacent nodes is L. n , the length of the automatic conveyor

[0268] For L c , its speed is v, so the time In the grid-based example, the automatic transport device travels in a straight line along the path in the four directions of east / south / west / north, and can only turn at the node. Turning requires deceleration / rotation / acceleration, and the whole process takes time t u In this way, assuming that the departure time of the automatic transport device is 0s, the total time T required to reach the target location is as shown in the following formula (7):

[0269]

[0270] Among them, n c Indicates the total number of nodes in the path, n u represents the number of turns in the path. Therefore, the optimal strategy selection criterion for the target path of the automatic transport device is: find an optimal path for the automatic transport device to reach the target location so that the total time T is minimized, that is,

[0271] Therefore, when there are multiple target paths finally determined, each target path can be evaluated by this method to select the optimal path.

[0272] In another example, for path planning of multiple automatic transport devices, multiple evaluation indicators can be constructed to obtain overall evaluation indicators of the multiple automatic transport devices based on these evaluation indicators.

[0273] Specifically, the evaluation criteria for path planning of multiple automatic transport devices may include: (a) path calculation time; (b) total displacement of the automatic transport device to complete a task (the automatic transport device travels from the starting position to the target position); (c) the number of tasks completed by the automatic transport device per unit time; (d) the total energy consumption of the automatic transport device to complete a task; (e) the number of turns made by the automatic transport device during driving. The evaluation criteria (a)-(e) can be comprehensively combined to transform the problem into a multi-objective optimization problem.

[0274] For example, suppose the i-th automatic transport device AGV iThe result of path planning is (i=1,2,…,n, n is the number of AGVs), where, Indicates AGV i The node number passed at time τ, Indicates AGV i Passing Node time, the path planning result of n AGVs is Ω={∏1,Π2,Π3,…,Π n As mentioned above, the time that the AGV travels on the adjacent nodes is t c , assuming that the energy consumption of the system when calculating the path is E1, the energy consumption of the AGV when driving in a straight line is E2, and the energy consumption when turning is E3, the evaluation indicators of multiple automatic transport device systems can be expressed by the following equations (8) to (7):

[0275]

[0276] Here, formula (8) represents the method of minimizing the system path calculation time CTR (for example, CTR can be directly obtained through program simulation, which can be obtained based on the computing performance of the device executing the algorithm, for example, it can be: the area or size of the calculated area × the unit computing time of the server); formula (9) represents the number of turns n in the driving path of the automatic transport device. u Minimize; The expression in formula (10) minimizes the total distance RDA traveled by the automatic transport device, N(∏) i Represents the path ∏ i The length (e.g. number of grids), L n represents the distance between every two adjacent nodes. For example, RDA can be directly obtained through program simulation. Formula (11) represents the method to minimize the total energy TEC consumption of the path. OnGoing(Δt τ ) represents the time period Δt τ The energy consumption of all automatic transport devices in the system; RTA in formula (12) represents minimizing the total time taken for the automatic transport device to travel from the starting position to the target position. The above formulas can satisfy the following conditions:

[0277]

[0278] Wherein, i=1,2,…,n, and n is the number of automatic transport devices.

[0279] According to the path planning method of the automatic transport device described in the embodiment of the present disclosure, the path optimization speed can be accelerated by limiting the leaf node expansion direction by comparing the positional relationship between the current position and the target position through heuristic search. In this method, the problem of low efficiency caused by the traditional Monte Carlo tree search after expanding to the target position and then performing back propagation can be avoided, and the back propagation is advanced, and the search tree is established by using a single-step update method.

[0280] In addition, this method also proposes a path evaluation criterion based on the path coefficient, which scores all feasible paths from the starting position to the target position to find the shortest path with the least turns.

[0281] In addition, this method can also introduce the idea of ​​dynamic programming multi-stage optimization problem into the search of transit positions, perform adaptive region segmentation according to the starting position and target position, and exclude nodes far away from the starting position / target position to increase path planning efficiency.

[0282] In addition, the path planning method for an automatic conveying device according to an embodiment of the present disclosure can also be extended to conflict-free path planning for multiple automatic conveying devices based on the path planning of a single automatic conveying device.

[0283] Specifically, this method proposes an adaptive multi-automatic transport device conflict-free path planning method based on path conflict critical sections: after obtaining all distance-time shortest paths based on the above-mentioned single automatic transport device, the distance-time shortest path of each automatic transport device is compared for conflict pre-detection, and then an adaptive method is used to obtain the final path of the automatic transport device and the schedule for reaching the node.

[0284] The following will refer to Figures 15 to 21 An example of using the path planning method according to an embodiment of the present disclosure is shown. For example, the example can be verified with reference to the above evaluation criteria.

[0285] Here, the driving speed of the automatic transport device is: uniform speed v = 0.5m / s; the driving time between two adjacent nodes is t c and the turning time t at the node u Both are 5s; loading and unloading time at the target location t w =30s; the energy consumption when the system performs path calculation is E1=800w, the energy consumption of the automatic transport device when driving in a straight line is E2=30w, and the energy consumption when parking and turning is E3=20w.

[0286] In the first example, seven automatic transport devices can be used to work simultaneously in a general assembly plant environment. Fig.15As shown, the circled numbers represent the workstations of the automatic conveyor, among which the circled bold numbers represent the charging stations of the automatic conveyor; the numbers without circles represent the intersections of the driving routes, and the dark bands where the driving routes are located represent the channels of the automatic conveyor. The starting positions and target positions of the seven automatic conveyors are as follows: AGV1: Workstation 6→Workstation 18; AGV2: Workstation 19→Workstation 8; AGV3: Workstation 12→Workstation 29; AGV4: Workstation 17→Workstation 14; AGV5: Workstation 26→Workstation 27; AGV6: Workstation 28→Workstation 22; AGV7: Workstation 1→Workstation 24. Fig.15 The result of rasterizing and coordinate conversion is as follows Fig.16 shown.

[0287] The algorithm proposed in the embodiment of the present disclosure is used to plan the distance-time shortest path from AGV1 to AGV7, and conflict detection is performed on its initial path. The results are as follows: there is a path conflict between AGV3 and AGV6, and the conflicting path segment Section = [517, 518, 519, 520, 521, 522, 523, 524, 525, 526, 527, 528, 529], with a length of len_sec = 13.

[0288] The solution to the above path conflict is as follows: AGV6 stops and waits for len_sec×t before entering the conflicting path segment Section. c =13×5=65s. The original paths of the 7 automatic conveyors are as follows: Fig.17A The path after conflict resolution is shown as Fig. 17B As shown, the black solid squares indicate the locations where path conflicts occur. The above-mentioned Algorithms 4 and 5 are used to perform conflict detection on the final shortest path to verify that there is no path conflict between the seven automatic transport devices.

[0289] The traditional dynamic programming (DP) and the algorithm proposed in this paper (DP-MCTS) are used to find the distance-time shortest path and resolve conflicts. They are run 1000 times to obtain the conflict-free shortest path. The average path calculation time is as follows: Fig.18 As shown in Figure 2, the average energy consumption is Fig.19 shown.

[0290] Specifically, the average path calculation time of the traditional DP algorithm is 0.8250s, and the average energy consumption is 1854.6295W; the average path planning time of the algorithm proposed in this paper is 0.0242s, and the average energy consumption is 1339.2579W. In this case, the path planning efficiency of the multi-automatic conveyor is improved by about 97.97%, and the system energy consumption is reduced by about 27.78%.

[0291] In the second example, the effectiveness in environments of different scales can be verified. In this example, the path calculation time is compared in environments with different numbers of nodes. As the scale of the environment (number of nodes) increases, the path calculation time of the traditional dynamic programming (DP) and the proposed algorithm (DP-MCTS) changes as shown in the following figure: Fig. 20 As shown in Figure 2, the energy consumption trend of the system is as follows: Fig.21 shown.

[0292] As shown in the above examples, in environments of different scales, the proposed algorithms have reduced path calculation time, saved system energy consumption, and improved system throughput to varying degrees. The larger the scale of the environment (the more nodes), the more obvious the advantages of the proposed algorithm.

[0293] According to the path planning method of the embodiment of the present disclosure, a dynamic conflict-free path planning algorithm for multiple automatic conveyors is proposed. The method rasterizes the environment according to the size of the automatic conveyor to achieve accurate positioning of the automatic conveyor, and proposes multiple improvements to the problem that the traditional Monte Carlo tree search path planning is inefficient.

[0294] Specifically, on the one hand, restriction links are added to guide the expansion direction of the search tree, giving the algorithm the ability of heuristic search; on the other hand, the single-step update method is used to evaluate the nodes to find a feasible path from the starting position to the target position; on the other hand, an evaluation standard based on the path coefficient is proposed to obtain the distance-time shortest path between the starting position and the target position; at the same time, the idea of ​​dynamic programming multi-stage optimization is combined with the above-mentioned search algorithm to realize the regional segmentation of the environment, narrow the path search range, and significantly improve the path search efficiency.

[0295] In addition, in order to solve the problem that multiple automatic conveying devices may cause path conflicts when they are running at the same time, an adaptive conflict resolution strategy based on conflict key segments is proposed, which can effectively solve the "potential" conflicts between the original shortest paths of multiple automatic conveying devices and the "temporary" conflicts in special circumstances. In addition, the automatic conveying device scheduling method based on distance indicators proposed in the embodiments of the present disclosure can reduce the idle time of multiple automatic conveying devices and increase the dynamics of the system.

[0296] For the above effects, example simulation verifies that this method can effectively improve the path planning efficiency of automatic transport devices in wind turbine assembly environments and reduce the energy consumption of the system.

[0297] According to a second aspect of an embodiment of the present disclosure, a path planning device for an automatic transport device is provided, such as Fig. 22 As shown, the path planning device may include a position determining unit 2210 , a searching unit 2220 , a planning unit 2230 and a path determining unit 2240 .

[0298] The position determination unit 2210 is configured to determine a starting position and a target position of the automatic transport device in a target environment.

[0299] The search unit 2220 is configured to take the starting position as the starting point and, according to a target search direction, search for the passing positions along the path from the starting position to the target position, wherein the target search direction is determined based on the positional relationship between the reference position and the target position.

[0300] The planning unit 2230 is configured to plan possible paths from the starting location to the target location based on the passing locations.

[0301] The path determination unit 2240 is configured to determine a target path of the automatic transport device from a starting position to a target position from possible paths.

[0302] As an example, the target environment is the assembly environment of a wind turbine generator set, and the target search direction includes: a direction pointing from a reference position to a target position, wherein the search unit 2230 is configured to search for the passing position in the following manner: starting from the starting position, searching for the passing position that satisfies the target search direction, wherein satisfying the target search direction means: the direction from the reference position to the passing position is the same as the target search direction or the same as the component of the target search direction.

[0303] As an example, the search unit 2230 is configured to: perform a search operation on an optional position in the target environment until the target position is searched, wherein the search unit 2230 is further configured to perform the search operation in the following manner: determine, from the optional positions in the target environment, a first position that satisfies an adjacency condition with the current reference position, wherein the adjacency condition includes: the first position is adjacent to the current reference position, and there is no obstacle between the first position and the current reference position; determine, from the first position, a second position that satisfies a direction condition with the current reference position, wherein the direction condition includes: the direction from the current reference position to the second position is the same as the current target search direction or the same as a component of the current target search direction; in response to the second position being different from the target position, determine the second position as a passing position, use the second position as a reference position for the next search operation, and perform the next search operation; in response to the second position being the same as the target position, end the search operation, wherein the reference position of the first search operation is the starting position.

[0304] As an example, there are multiple possible paths, wherein the path determination unit 2250 is configured to: determine the path coefficient of each possible path, wherein the path coefficient represents: the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; determine the possible path with the smallest path coefficient as the target path.

[0305] As an example, the path determination unit 2250 is configured to determine that the automatic transport device experiences a turn by: in response to an angular difference between the direction of travel of the automatic transport device when it moves to the current position and the direction of travel of the automatic transport device when it moves to the next position, determining that the automatic transport device experiences a turn when it moves from the current position to the next position.

[0306] As an example, the target environment is an assembly environment of a wind turbine generator set, wherein the path determination unit 2250 is configured to: determine a distance coefficient and a turning coefficient respectively according to the distance traveled by the automatic transport device from the starting position to the target position via a possible path and the number of turns experienced; determine a turning weight according to the area size of the assembly environment, wherein the turning weight is positively correlated with the area size; weight the turning coefficient using the turning weight to obtain a weighted turning coefficient; and obtain a path coefficient based on the distance coefficient and the weighted turning coefficient.

[0307] As an example, the search unit 2230 is configured to: divide a target area in the target environment according to the starting position, the target position and the environmental boundary of the target environment, wherein the target area includes the starting position and the target position; and search for the passing position in the target area according to the target search direction.

[0308] As an example, the search unit 2230 is configured to determine the target area in the following manner: determine a first candidate position among the starting position and the target position that is closest to the environment boundary in a preset direction; determine a second candidate position at a preset distance from the first candidate position along the preset direction; and determine the boundary of the target area in the preset direction based on the positional relationship between the second candidate position and the environment boundary to determine the target area.

[0309] As an example, there are multiple automatic transport devices, each of which corresponds to its own target position, wherein the path planning device also includes a path updating unit, and the path updating unit is configured to: determine the initial path of each automatic transport device according to the number of target paths corresponding to each automatic transport device; in response to the existence of conflicting path segments between the initial paths of each automatic transport device, update the initial paths of the conflicting automatic transport devices by performing a path updating operation until there is no conflict between the initial paths of each automatic transport device, thereby obtaining the final paths of each automatic transport device, wherein the conflict between the target paths of the automatic transport devices means that the automatic transport devices will be in the same path segment at the same time when traveling according to their respective target paths; in response to the absence of conflicting path segments between the initial paths of each automatic transport device, use the current initial path of each automatic transport device as the final path of each automatic transport device.

[0310] As an example, the path updating unit is configured to update the initial path of the conflicting automatic transport devices in the following manner: by adjusting the time when at least one of the conflicting target transport devices reaches the conflicting path segment, the initial path of the at least one target transport device is updated.

[0311] As an example, the target environment is an assembly environment of a wind turbine generator set, wherein the path updating unit is configured to update the initial path of the at least one target transport device in the following manner: determining a distance index between the starting position and the target position of each of the conflicting target transport devices, the distance index being used to characterize the distance between the starting position and the target position; for any two conflicting target transport devices, determining the initial paths of the two conflicting target transport devices by performing the following operations: in response to the target position of the first target transport device being located on the initial path of the second target transport device, determining the waiting time of the second target transport device before entering the conflicting path segment based on the length of the conflicting path segment between the first target transport device and the second target transport device, the moving speed of the first target transport device, and a preset stay time, so as to update the initial path of the second target transport device; in response to the target position of the first target transport device not being located on the initial path of the second target transport device, determining the waiting time of the second target transport device before entering the conflicting path segment based on the length of the conflicting path segment between the first target transport device and the second target transport device and the moving speed of the first target transport device, so as to update the initial path of the second target transport device, wherein the distance index of the first target transport device is greater than the distance index of the second target transport device.

[0312] As an example, the path update unit is configured to determine the initial path of each automatic conveying device in the following manner: in response to the number of target paths of the first automatic conveying device being less than the number of target paths of the second automatic conveying device, taking the target path of the second automatic conveying device that does not overlap with the initial path of the first automatic conveying device as the initial path of the second automatic conveying device; in response to the number of target paths of the first automatic conveying device being equal to the number of target paths of the second automatic conveying device, determining the initial paths of the first automatic conveying device and the second automatic conveying device from their respective target paths, so that there is no overlap between the initial paths of the first automatic conveying device and the second automatic conveying device.

[0313] Regarding the device in the above embodiment, the specific manner in which each unit performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0314] The path planning method for an automatic transport device according to an embodiment of the present disclosure can be executed by a computer device, which may include: a processor; a memory for storing processor executable instructions, wherein the processor executable instructions, when executed by the processor, prompt the processor to execute the path planning method for an automatic transport device according to an embodiment of the present disclosure.

[0315] As an example, the computer device can be a PC, a tablet device, a personal digital assistant, a smart phone, or other device capable of executing the above-mentioned instruction set. Here, the computer device is not necessarily a single electronic device, but can also be any device or circuit that can execute the above-mentioned instructions (or instruction sets) individually or jointly. The computer device can also be part of an integrated control system or system manager, or can be configured as a portable electronic device that is interconnected with a local or remote (e.g., via wireless transmission) interface.

[0316] In a computer device, a processor may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller or a microprocessor. As an example and not limitation, a processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0317] The processor can execute instructions or codes stored in the memory, wherein the memory can also store data. Instructions and data can also be sent and received through the network via the network interface device, wherein the network interface device can adopt any known transmission protocol.

[0318] The memory may be integrated with the processor, for example, RAM or flash memory is arranged within an integrated circuit microprocessor or the like. In addition, the memory may include a separate device, such as an external disk drive, a storage array, or any other storage device that can be used by a database system. The memory and the processor may be operatively coupled, or may communicate with each other, such as through an I / O port, a network connection, etc., so that the processor can read files stored in the memory.

[0319] In addition, the computer device may also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.) All components of the computer device may be connected to each other via a bus and / or a network.

[0320] According to an embodiment of the present disclosure, a computer-readable storage medium may also be provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute the path planning method for an automatic transport device according to an embodiment of the present disclosure.

[0321] Specifically, the path planning method for an automatic transport device according to an embodiment of the present disclosure can be written as a computer program and stored on a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by at least one processor, the at least one processor is prompted to execute the path planning method for an automatic transport device according to the exemplary embodiment of the present disclosure. Examples of computer-readable storage media include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer program and any associated data, data files and data structures in a non-transitory manner and provide the computer program and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. In one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0322] According to an embodiment of the present disclosure, a computer program product may also be provided, which includes computer executable instructions. When the computer executable instructions are executed by at least one processor, the path planning method for the automatic transport device according to the embodiment of the present disclosure is implemented.

[0323] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0324] In addition, it should be noted that although several examples of each step are described above with reference to specific drawings, it should be understood that the embodiments of the present disclosure are not limited to the combinations given in the examples, and the steps appearing in different drawings may be combined, and the order of execution of the steps may be changed, which is not exhaustive here.

[0325] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

[0326] The specific implementation methods of the present disclosure have been described in detail above. Although some embodiments have been shown and described, those skilled in the art should understand that these embodiments may be modified and varied without departing from the principles and spirit of the present disclosure, the scope of which is defined by the claims and their equivalents. These modifications and variations should also be within the scope of protection of the claims of the present disclosure.

Claims

1. A path planning method for an automatic transport device, characterized in that: The path planning method comprises: Determining a starting position and a target position of the automatic transport device in a target environment; Taking the starting position as a starting point, searching for the passing positions of the path from the starting position to the target position according to a target search direction, wherein the target search direction is determined based on the positional relationship between a reference position and the target position; Based on the passing positions, planning possible paths from the starting position to the target position; From the possible paths, a target path of the automatic transport device from the starting position to the target position is determined.

2. The path planning method according to claim 1, characterized in that: The target environment is the assembly environment of a wind turbine generator set, and the target search direction includes: a direction from the reference position to the target position, The passing location is searched in the following manner: Taking the starting position as the starting point, searching for the passing position that satisfies the target search direction, Among them, satisfying the target search direction means that: the direction from the reference position to the passing position is the same as the target search direction or the same as the component of the target search direction.

3. The path planning method according to claim 2, characterized in that: The step of searching the passing position that satisfies the target search direction with the starting position as the starting point includes: performing a search operation on an optional position in the target environment until the target position is found. The search operation is performed in the following manner: Determine, from the optional positions in the target environment, a first position that satisfies an adjacency condition with the current reference position, wherein the adjacency condition includes: the first position is adjacent to the current reference position, and there is no obstacle between the first position and the current reference position; Determine, from the first position, a second position that satisfies a direction condition with respect to the current reference position, wherein the direction condition includes: a direction from the current reference position to the second position is the same as the current target search direction or is the same as a component of the current target search direction; In response to the second position being different from the target position, determining the second position as the passing position, using the second position as a reference position for a next search operation, and performing the next search operation, In response to the second position being the same as the target position, the search operation is terminated, wherein the reference position of the first search operation is the starting position.

4. The path planning method according to claim 1, characterized in that: There are multiple possible paths. Wherein, determining the target path of the automatic transport device from the starting position to the target position from the possible paths includes: Determine a path coefficient for each possible path, wherein the path coefficient represents: the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; The possible path with the smallest path coefficient is determined as the target path.

5. The path planning method according to claim 4, characterized in that: The automatic transport device is determined to have experienced a turn by: In response to an angular difference between a direction in which the automatic conveyor moves to the current position and a direction in which the automatic conveyor moves to the next position, it is determined that the automatic conveyor moves from the current position to the next position through a turn.

6. The path planning method according to claim 4 or 5, characterized in that: The target environment is an assembly environment of a wind turbine generator set, wherein determining the path coefficient of each possible path includes: Determine a distance coefficient and a turning coefficient according to the distance traveled by the automatic transport device from the starting position to the target position via the possible path and the number of turns experienced; Determining a turning weight according to the area size of the final assembly environment, wherein the turning weight is positively correlated with the area size; The turning coefficient is weighted by using the turning weight to obtain a weighted turning coefficient; The path coefficient is obtained based on the distance coefficient and the weighted turning coefficient.

7. The path planning method according to claim 1, characterized in that: The step of searching for the passing positions of the path from the starting position to the target position according to the target search direction includes: According to the starting position, the target position and the environment boundary of the target environment, a target area is divided in the target environment, wherein the target area includes the starting position and the target position; According to the target search direction, the passing position is searched in the target area.

8. The path planning method according to claim 7, characterized in that: The target area is determined by: Determine a first candidate position, among the starting position and the target position, which is closest to the environment boundary in a preset direction; Determine a second candidate position at a preset distance from the first candidate position along the preset direction; Based on the positional relationship between the second candidate position and the environment boundary, a boundary of the target area in the preset direction is determined to determine the target area.

9. The path planning method according to claim 1, characterized in that: There are multiple automatic transport devices, each of which corresponds to its own target location. Wherein, the path planning method further includes: Determine the initial path of each automatic transport device according to the number of target paths corresponding to each automatic transport device; In response to the existence of conflicting path segments between the initial paths of the automatic transport devices, the initial paths of the conflicting automatic transport devices are updated by executing a path update operation until there is no conflict between the initial paths of the automatic transport devices, thereby obtaining final paths of the automatic transport devices, wherein the conflict between the target paths of the automatic transport devices means that the automatic transport devices will be in the same path segment at the same time when traveling according to their respective target paths; In response to the absence of conflicting path segments between the initial paths of the automatic conveying devices, the current initial path of each automatic conveying device is used as the final path of each automatic conveying device.

10. The path planning method according to claim 9, characterized in that: Update the initial path of the conflicting automatic transport device by: The initial path of the at least one target transport device is updated by adjusting the time when at least one of the conflicting target transport devices arrives at the conflicting path segment.

11. The path planning method according to claim 10, characterized in that: The target environment is an assembly environment of a wind turbine generator set, wherein the initial path of the at least one target transport device is updated in the following manner: Determine a distance index between the starting position and the target position of each of the conflicting target transport devices, wherein the distance index is used to represent the distance between the starting position and the target position; For any two conflicting target transport devices, the initial paths of the two conflicting target transport devices are determined by performing the following operations: In response to the target position of the first target transport device being located on the initial path of the second target transport device, based on the length of the conflicting path segment between the first target transport device and the second target transport device, the moving speed of the first target transport device, and a preset dwell time, determining a waiting time for the second target transport device before entering the conflicting path segment, so as to update the initial path of the second target transport device; In response to the target position of the first target transport device not being located on the initial path of the second target transport device, based on the length of the conflicting path segment between the first target transport device and the second target transport device and the moving speed of the first target transport device, determining a waiting time for the second target transport device before entering the conflicting path segment to update the initial path of the second target transport device, Wherein, the distance index of the first target transport device is greater than the distance index of the second target transport device.

12. The path planning method according to claim 9, characterized in that: The initial path of each automatic transport device is determined by: In response to the number of target paths of the first automatic conveyor being less than the number of target paths of the second automatic conveyor, taking a target path among the target paths of the second automatic conveyor that does not overlap with an initial path of the first automatic conveyor as an initial path of the second automatic conveyor; In response to the number of target paths of the first automatic conveying device being equal to the number of target paths of the second automatic conveying device, initial paths of the first automatic conveying device and the second automatic conveying device are determined from their respective target paths, so that there is no overlap between the initial paths of the first automatic conveying device and the second automatic conveying device.

13. A path planning device for an automatic transport device, characterized in that: The path planning device comprises: a position determination unit, configured to determine a starting position and a target position of the automatic transport device in a target environment; A search unit is configured to search for a path from the starting position to the target position according to a target search direction, starting from the starting position, wherein the target search direction is determined based on a positional relationship between a reference position and the target position; A planning unit, configured to plan a possible path from the starting position to the target position based on the passing position; The path determination unit is configured to determine a target path of the automatic transport device from the starting position to the target position from the possible paths.

Images