An intelligent arrangement method of AGV garage parking spaces based on graph structure

Abstract

The present invention discloses an intelligent arrangement method of parking spaces in an AGV garage based on a graph structure, and relates to the technical field of parking space arrangement. The method comprises: structured processing of garage drawings to form column grid units; constructing a column grid graph structure based on the column grid units, traversing adjacent column grid units in the column grid graph structure using a breadth-first algorithm and assigning serial numbers in sequence; expanding and connecting column grid units according to the traversal order of random directions to form several partition division schemes, and obtaining multiple sequential partition schemes, arranging parking spaces in the partitions in sequence, and determining the optimal parking space arrangement scheme; generating a single channel using a dynamic programming algorithm, and generating a full-map lane according to a backtracking algorithm; allocating single- and double-layer parking spaces, and converting lane parking spaces, and selecting the optimal scheme based on the statistical results of the number of lane parking spaces, single- and double-layer parking spaces, and the total number of parking spaces. The present invention can improve the efficiency and quality of the parking space arrangement design of AGV garages in large areas and complex spatial environments.

Description

Technical Field

[0001] The present invention relates to the technical field of parking space arrangement, and in particular to an intelligent arrangement method of AGV garage parking spaces based on a graph structure. Background Art

[0002] In recent years, with the acceleration of urbanization and the sharp increase in the number of motor vehicles, parking difficulties have become an important problem that plagues urban management and residents' daily lives. Traditional manual parking methods not only occupy a large amount of space resources, but also have many shortcomings such as low efficiency, high management costs and safety hazards. In order to meet this challenge, the automatic guided vehicle (AGV) parking garage has emerged as a new parking technology. Through the intelligent vehicle dispatching and management system, the AGV parking garage can achieve efficient and convenient parking services, greatly improving the utilization efficiency and management level of the parking lot; compared with traditional manual parking, the AGV parking garage can not only save a lot of human resources, but also maximize the use of limited parking space by optimizing parking space layout and path planning to meet the growing parking demand; in addition, with the continuous development of advanced technologies such as the Internet of Things, big data and artificial intelligence, the intelligence and automation level of the AGV parking garage are also constantly improving, providing important infrastructure support for the construction of future smart cities.

[0003] Existing AGV parking technology mainly relies on the manual design of designers to arrange parking spaces and plan paths. In this mode, manual arrangement needs to comprehensively consider multiple factors such as the area of ​​the garage, the spatial environment, the movement characteristics of the vehicle, and the layout of various parking spaces in order to formulate a reasonable parking plan. AGV parking spaces are usually densely laid, and traffic problems are solved internally through frames with different attributes, including single-upper-layer laneway parking spaces, double-layer parking spaces that can park both layers, and single-layer parking spaces that can only be used as a single bottom layer due to insufficient height. This diversified parking space layout method makes the AGV parking garage highly flexible and adaptable in design; however, with the expansion of the scale of the garage and the complexity of the spatial environment, the manual arrangement method has gradually exposed its inevitable limitations. Manual arrangement requires a lot of time and energy to be repeatedly adjusted and optimized, and in the face of a complex spatial environment, it is often difficult to find the optimal parking space arrangement plan, which affects the operating efficiency and user experience of the entire parking system.

[0004] The existing manual arrangement method has significant shortcomings when dealing with AGV parking garages with large areas and complex spatial environments: first, as the scale of the garage expands, the factors that need to be considered in manual arrangement increase exponentially, causing the design process to become extremely cumbersome and time-consuming, and it is difficult to meet the actual needs of rapid development; second, when dealing with complex spatial environments, manual arrangement is prone to unsatisfactory or even wrong arrangement results, resulting in reduced parking space utilization and low parking efficiency; in addition, manual arrangement is highly subjective, and the quality and effect of parking space design are easily limited by the designer's personal experience and skill level, making it difficult to ensure consistent and high-quality design output; more importantly, with the continuous evolution of AGV technology and parking needs, traditional manual arrangement methods are difficult to quickly adapt to and respond to new technology and demand changes, limiting the further development of AGV parking garages in terms of intelligence and automation. Based on the above problems, a more efficient and intelligent AGV parking space arrangement method is urgently needed.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the shortcomings of the prior art, the present invention provides an intelligent arrangement method for AGV garage parking spaces based on a graph structure, which has the advantages of improving design efficiency, optimizing parking space layout, ensuring efficient operation of the parking system and a good user experience, thereby solving the problems of slow speed, easy errors and poor effect in the prior art of AGV garage parking space arrangement design in large-area and complex space environments.

[0008] (II) Technical solution

[0009] In order to achieve the above advantages of improving design efficiency, optimizing parking space layout, ensuring efficient operation of the parking system and good user experience, the specific technical solutions adopted by the present invention are as follows:

[0010] A method for intelligently arranging parking spaces in an AGV garage based on a graph structure comprises the following steps:

[0011] S1. Structural processing of garage drawings, and extraction of geometric information layers. Based on the geometric information layers, processing of axis grid and column drawings to form column grid units;

[0012] S2. construct a column grid structure based on the column grid units, traverse the adjacent column grid units in the column grid structure using a breadth-first algorithm, and assign column grid unit numbers in sequence according to the traversal order;

[0013] S3. Expand and connect the column grid units according to the traversal order of the random direction to form several partition division schemes, and obtain multiple sequential partition division schemes based on the random arrangement of different orders under each partition division scheme, arrange the parking spaces in the partitions in order, and determine the optimal parking space arrangement scheme;

[0014] S4. Based on the optimal parking space arrangement plan, form the lane parking space arrangement, and use the dynamic programming algorithm to generate a single channel, generate the full map lane according to the backtracking algorithm, and determine all parking space types;

[0015] S5. Allocate single-layer parking spaces and double-layer parking spaces according to the parking space type and the result of geometric intersection operation, and convert the lane parking spaces. Select the optimal solution based on the statistical results of the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces.

[0016] Furthermore, the garage drawings are processed in a structured manner and the geometric information layer is extracted. According to the geometric information layer, the axis grid and column drawings are processed to form the column grid unit including:

[0017] S11. Import CAD drawings, separate the elements in the drawings into different layers, extract the geometric information in the layers, and convert it into structured data. The geometric information includes the axis grid and the position of the columns.

[0018] S12, based on the extracted geometric information, identifying the axis grid lines in the layer and determining the axis grid intersection points;

[0019] S13, based on the extracted geometric information, identifying the column positions in the layer, and corresponding them to the axis grid intersections to form the boundaries of the column grid units;

[0020] S14. Select the intersection of the axis grid as the center point of the column grid unit, and divide the column grid unit into independent units based on the boundary of the column grid unit.

[0021] Furthermore, the column grid graph structure is an undirected graph structure, each node in the column grid graph structure represents a column grid unit, and each edge in the column grid graph structure represents a connection relationship between column grid units.

[0022] Furthermore, constructing a column grid structure based on the column grid units, traversing adjacent column grid units in the column grid structure using a breadth-first algorithm, and assigning column grid unit numbers in sequence include the following steps:

[0023] S21. Based on the column grid structure, select a column grid unit whose center is closest to the center of the parking room as a starting node, and find a column grid unit that shares a common edge with the starting node to obtain a number of adjacent column grid units.

[0024] S22, randomly selecting a column grid unit from adjacent column grid units as the next starting node;

[0025] S23, using a breadth-first algorithm, traverse the adjacent column grid units from the starting node, and if the traversal reaches a dead end, use a backtracking algorithm to return to the previous layer of multiple-selection nodes and continue traversing;

[0026] S24. Repeat steps S22 to S23, and assign column grid unit numbers in order, traverse all nodes in the graph structure layer by layer from near to far, and obtain the traversal order of each node.

[0027] Furthermore, according to the traversal order of the random direction, the connected column grid units are expanded to form several partitioning schemes, and based on the random arrangement of different orders under each partitioning scheme, multiple sequential partitioning schemes are obtained, and the parking spaces in the partitions are arranged in order. Determining the optimal parking space arrangement scheme includes:

[0028] S31, based on the traversal order of each node, randomly select a direction from the smallest column grid unit to expand the connection, form a partition, and determine several partition division schemes;

[0029] S32. According to each partition division scheme, select the partition closest to the parking room, and randomly arrange the shared edge partitions in order to obtain several sequential partition schemes with different orders;

[0030] S33. Based on the sequential partitioning schemes of different orders, the parking spaces in the partitions are arranged in order, and the maximum number of parking spaces is taken as the goal, and the arrangement feature points of the previous partition are transferred to determine the optimal parking space arrangement scheme.

[0031] Furthermore, based on the traversal order of each node, a direction is randomly selected from the smallest column grid unit for expansion connection to form a partition, and several partition division schemes are determined, including the following steps:

[0032] S311, based on the traversal order of each node, randomly select one-way or two-way, and expand the connection with the adjacent column grid units starting from the column grid unit with the smallest sequence number according to the selected direction to form a partition;

[0033] S312, expanding the connection again according to the smallest serial number of the column grid units remaining after each expansion and connection, until all column grid units are connected into partitions;

[0034] S313. Based on the maximum number of generated solutions, repeat steps S31 and S32 to obtain several partition division solutions.

[0035] Further, the arranged feature points include boundary points, original feature points and projected feature points;

[0036] The boundary points include each corner point of the parking space shape boundary and the intersection points on the vertical auxiliary lines formed in the partition transfer direction;

[0037] The original feature points include any point of any parking space in the partition of the previous order;

[0038] The projected feature points include points where the current feature points are projected onto the auxiliary line of the next partition along the partition transfer direction.

[0039] Furthermore, based on the optimal parking space arrangement plan, the lane parking space arrangement is formed, and a single channel is generated using a dynamic programming algorithm, and a full-map lane is generated according to a backtracking algorithm. Determining all parking space types includes the following steps:

[0040] S41, based on the optimal parking space arrangement plan, forming a lane parking space arrangement and establishing a parking space map structure;

[0041] S42, based on the parking space map structure, using dynamic programming algorithm to generate a single channel;

[0042] S43, using a backtracking algorithm, backtracking the last multiple-selection node of a single lane, and selecting an unselected direction to generate a new channel;

[0043] S44, repeat step S43 until the full map lanes are generated and all parking space types are determined, including channel parking spaces and non-channel parking spaces.

[0044] Furthermore, based on the parking space map structure, a single channel is generated using a dynamic programming algorithm including:

[0045] S421, starting to arrange parking spaces based on the parking space map structure, taking the parking house as a starting point;

[0046] S422, after the previous parking space is generated, calculate the potential position score of the next lane parking space, and select the parking space node with the highest score to extend the lane;

[0047] The expression for calculating the potential position score of the next channel parking space is:

[0048]

[0049] In the formula, P represents the score of the potential position of the next channel parking space; N represents the basic score;

[0050] S423, continuously extending the lane according to the scores of the parking spaces that have been passed and the potential parking spaces, converting the original parking spaces into lane parking spaces, until the passage reaches the parking spaces at the edge of the parking lot or the number of potential parking spaces in each next passage is less than or equal to one, thus completing the generation of a single passage.

[0051] Further, according to the parking space type and in combination with the result of geometric intersection operation, single-layer parking spaces and double-layer parking spaces are allocated, and laneway parking spaces are converted. Based on the statistical results of the number of laneway parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces, the optimal solution is selected, including the following steps:

[0052] S51, with the goal of maximizing the number of parking spaces, allocating double-decker parking space attributes to all non-aisle parking spaces, and alleyway parking space attributes to all aisle parking spaces;

[0053] S52, setting the geometric boundary of the low-rise area, and performing geometric intersection operation with all generated parking spaces, if the generated parking space intersects with the low-rise area, changing the parking space attribute;

[0054] Changing the parking space attributes includes: if it was originally a double-layer parking space, it will be converted into a single-layer parking space; if it was originally an alley parking space, it will be converted into an empty space without a parking space;

[0055] S53. Count the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces, select the optimal solution based on the statistical results and export it.

[0056] (III) Beneficial effects

[0057] Compared with the prior art, the present invention provides an intelligent arrangement method of AGV garage parking spaces based on a graph structure, which has the following beneficial effects:

[0058] (1) The present invention fully automates the design process by introducing advanced graph structures and intelligent algorithms. Only basic design boundary conditions need to be input to generate the optimal parking space layout plan in a short time. This automated design process not only greatly shortens the design cycle, but also reduces human errors, ensuring the accuracy and reliability of the design plan. In addition, the present invention can complete a layout plan containing more than 20 partitions and more than 200 AGV parking spaces within 2 minutes. This efficient processing capability provides strong technical support for the design of large-scale AGV garages. The fast design speed means that the early preparation work of the project can be completed faster, thereby accelerating the overall progress of the project, reducing project costs, and better meeting the needs of different projects.

[0059] (2) The present invention proposes a more universal and adaptable arrangement algorithm and a complete workflow, which can completely arrange the parking spaces in the AGV garage. This innovation not only improves the flexibility and applicability of the design, but also provides a guarantee for the efficient operation of the parking system and the good user experience. The present invention can flexibly cope with various complex garage environments through graph structure and dynamic programming algorithm, and can generate the optimal parking space arrangement plan whether it is a multi-storey garage, underground garage or open-air garage. In addition, the present invention ensures the efficient operation of the AGV car by optimizing the layout of parking spaces and passages. Reasonable parking space arrangement can reduce the waiting time and driving distance of the AGV car during the process of picking up and parking, thereby improving the overall efficiency of the parking system. This is particularly important for improving the user experience. Users can complete parking and picking up operations faster and enjoy more convenient and comfortable parking services, which significantly improves the design and operation level of the AGV garage. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 It is a flowchart of a method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to an embodiment of the present invention;

[0062] Figure 2 is a schematic diagram of a column grid unit in a graph-structure-based intelligent arrangement method for AGV garage parking spaces according to an embodiment of the present invention;

[0063] Figure 3 is a schematic diagram of finding adjacent column grid units in a graph-structure-based intelligent arrangement method for AGV garage parking spaces according to an embodiment of the present invention;

[0064] Figure 4 is a schematic diagram of randomly selecting column grid units in a graph-structure-based intelligent arrangement method for AGV garage parking spaces according to an embodiment of the present invention;

[0065] Figure 5 is a schematic diagram of the traversal order of nodes in the graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0066] Figure 6 It is a schematic diagram of partitions formed by extending connections in a method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to an embodiment of the present invention;

[0067] Figure 7is a schematic diagram of a partition division scheme in a graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0068] Figure 8 It is a schematic diagram of the arrangement of multiple sequential partitioning schemes of different orders in the graph-structure-based intelligent arrangement method for parking spaces in an AGV garage according to an embodiment of the present invention;

[0069] Fig. 9 It is a schematic diagram of front and rear partitions transferring shared parking space features to maintain an aligned arrangement state in a graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0070] Fig.10 It is a schematic diagram of the arrangement of multiple partitioned parking spaces in the graph-structure-based intelligent arrangement method for AGV garage parking spaces according to an embodiment of the present invention;

[0071] Fig.11 It is a schematic diagram of parking space node connection and edge attribute setting in a graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0072] Fig.12 It is a schematic diagram of the connection structure of the whole parking space graph in the method for intelligently arranging parking spaces of an AGV garage based on a graph structure according to an embodiment of the present invention;

[0073] Fig.13 It is a schematic diagram of calculating the score of a channel generated by a dynamic programming algorithm in a method for intelligently arranging parking spaces of an AGV garage based on a graph structure according to an embodiment of the present invention;

[0074] Fig.14 is a schematic diagram of calculating the score of another channel generated by a dynamic programming algorithm in a method for intelligently arranging parking spaces of an AGV garage based on a graph structure according to an embodiment of the present invention;

[0075] Fig.15 It is a schematic diagram of dynamic process planning in a method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to an embodiment of the present invention;

[0076] Fig.16 is a schematic diagram of a backtracking algorithm in a graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0077] Fig.17 It is a schematic diagram of a full-graph channel after multiple backtrackings in a graph-structure-based AGV garage parking space intelligent arrangement method according to an embodiment of the present invention;

[0078] Fig.18 It is a schematic diagram of scheme visualization and number statistics of different types of parking spaces in a method for intelligently arranging parking spaces of an AGV garage based on a graph structure according to an embodiment of the present invention;

[0079] Fig.19 It is a schematic diagram of geometric intersection operations of low-rise and high-rise areas in a graph-structure-based intelligent arrangement method for AGV garage parking spaces according to an embodiment of the present invention. DETAILED DESCRIPTION

[0080] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0081] According to an embodiment of the present invention, a method for intelligently arranging parking spaces in an AGV garage based on a graph structure is provided.

[0082] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an embodiment of the present invention, the method for intelligently arranging parking spaces in an AGV garage based on a graph structure includes the following steps:

[0083] S1. Structural processing of garage drawings, and extraction of geometric information layers. Based on the geometric information layers, processing of axis grid and column drawings to form column grid units;

[0084] S2. construct a column grid structure based on the column grid units, traverse the adjacent column grid units in the column grid structure using a breadth-first algorithm, and assign column grid unit numbers in sequence according to the traversal order;

[0085] S3. Expand and connect the column grid units according to the traversal order of the random direction to form several partition division schemes, and obtain multiple sequential partition division schemes based on the random arrangement of different orders under each partition division scheme, arrange the parking spaces in the partitions in order, and determine the optimal parking space arrangement scheme;

[0086] S4. Based on the optimal parking space arrangement plan, form the lane parking space arrangement, and use the dynamic programming algorithm to generate a single channel, generate the full map lane according to the backtracking algorithm, and determine all parking space types;

[0087] S5. Allocate single-layer parking spaces and double-layer parking spaces according to the parking space type and the result of geometric intersection operation, and convert the lane parking spaces. Select the optimal solution based on the statistical results of the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces.

[0088] In one embodiment, the garage drawings are structured and processed, and a geometric information layer is extracted. According to the geometric information layer, the axis grid and column drawings are processed to form a column grid unit, which includes:

[0089] S11. Import CAD drawings, separate the elements in the drawings into different layers, extract the geometric information in the layers, and convert it into structured data. The geometric information includes axis grid lines and column positions.

[0090] S12, based on the extracted geometric information, identifying the axis grid lines in the layer and determining the axis grid intersection points;

[0091] S13, based on the extracted geometric information, identifying the column positions in the layer, and corresponding them to the axis grid intersections to form the boundaries of the column grid units;

[0092] S14. Select the intersection of the axis grid as the center point of the column grid unit, and divide the column grid unit into independent units based on the boundary of the column grid unit.

[0093] Specifically, ① structural processing of the CAD drawings of the AGV garage; ② importing CAD drawings; ③ importing and real-time modification of geometric information layers; ④ forming separate column grid units by processing the axis grid and column drawings. The results are as follows Figure 2 shown.

[0094] In one embodiment, the column grid graph structure is an undirected graph structure, each node in the column grid graph structure represents a column grid unit, and each edge in the column grid graph structure represents a connection relationship between column grid units.

[0095] In one embodiment, constructing a column grid structure based on column grid units, traversing adjacent column grid units in the column grid structure using a breadth-first algorithm, and assigning column grid unit serial numbers in sequence include the following steps:

[0096] S21. Based on the column grid structure, select a column grid unit whose center is closest to the center of the parking room as a starting node, and find a column grid unit that shares a common edge with the starting node to obtain a number of adjacent column grid units.

[0097] S22, randomly selecting a column grid unit from adjacent column grid units as the next starting node;

[0098] S23, using a breadth-first algorithm, traverse the adjacent column grid units starting from the starting node, and if the traversal reaches a dead end, use a backtracking algorithm to return to the previous layer of multiple-selection nodes and continue traversing;

[0099] S24. Repeat steps S22 to S23, and assign column grid unit numbers in order, traverse all nodes in the graph structure layer by layer from near to far, and obtain the traversal order of each node.

[0100] It should be noted that the Breadth-First Search (BFS) algorithm is an algorithm used for graph traversal or search. Its basic principle is that when starting from the starting node, it first visits all nodes directly connected to the starting node, and then visits the unvisited neighbor nodes of these nodes in turn, and so on, until all nodes are visited.

[0101] Specifically, Figure 3 As shown, ⑤ a column grid graph structure (Graph) is formed between column grid units, and the adjacent column grid units are traversed using the breadth-first algorithm, and column grid numbers are assigned in sequence; starting from a column grid unit that is closest to the parking room (the distance between the center of the column grid unit and the center of the parking room), several adjacent column grid units are found through the common edge judgment of the unit boundaries.

[0102] Specifically, a unit adjacent to it is randomly selected as the next unit. The random selection result is as follows: Figure 4 shown.

[0103] Specifically, in the column grid structure, each grid unit is represented as a node, and the connection relationship between nodes is represented by edges. In this way, the geometric space unit is converted into an undirected graph structure (Graph), and then the nodes in the graph are traversed using the breadth-first algorithm starting from the node closest to the parking room, and the nodes of the graph are traversed layer by layer from near to far, and the backtracking algorithm is used to return to the previous layer of multiple selection nodes when a dead end is reached, so as to ensure that every node in the entire graph is traversed, and finally the traversal order of each node is given. The traversal result is as follows: Figure 5 shown.

[0104] In one embodiment, the column grid units are connected and expanded according to the traversal order of the random direction to form several partitioning schemes, and based on the random arrangement of different orders under each partitioning scheme, multiple sequential partitioning schemes are obtained, and the parking spaces in the partitions are arranged in order. Determining the optimal parking space arrangement scheme includes:

[0105] S31, based on the traversal order of each node, randomly select a direction from the smallest column grid unit to expand the connection, form a partition, and determine several partition division schemes;

[0106] Specifically, ⑥ a partition that can accommodate more parking spaces is formed through single and bidirectional traversal order column network connections.

[0107] Specifically, since more parking spaces may be obtained by connecting column grid units in one direction to form partitions, in this embodiment:

[0108] 1) If the unit parking space size is a*b;

[0109] 2) Take the most common rectangular partition i as an example;

[0110] 3) Length and width: X i *a+C i and Y i *b+D i (where C and D are less than a and b respectively);

[0111] 4) The maximum number of parking spaces is X i *Y i ;

[0112] 5) If the column grid units are connected in n directions to form a large partition;

[0113] 6) The length and width of the complete partition are: and Y i *b+D i ,and Therefore, after forming a large partition, the number of parking spaces arranged inside is greater than or equal to

[0114] S32. According to each partition division scheme, select the partition closest to the parking room, and randomly arrange the shared edge partitions in order to obtain several sequential partition schemes with different orders;

[0115] Specifically, after step S31, a partitioning scheme is obtained. ⑦ Then, similar partition units are transformed into graph structures, and the breadth-first algorithm and backtracking algorithm are combined to distribute the order (it can be understood that the large partition is regarded as a number of column grid units). The order of the common edge partitions is randomly arranged from the large partition closest to the parking room to obtain multiple sequential partitioning schemes with different orders, such as Figure 8 As shown, Figure (a) and Figure (b) respectively show two different orders of sequential partitioning scheme arrangements, and the internal numbers in the figures represent the order of the partitioning scheme arrangement.

[0116] S33. Based on the sequential partitioning schemes of different orders, the parking spaces in the partitions are arranged in order, and the maximum number of parking spaces is taken as the goal, and the arrangement feature points of the previous partition are transferred to determine the optimal parking space arrangement scheme.

[0117] Specifically, after forming multiple sequential partitioning schemes, ⑧ arrange the parking spaces in the partitions in order and transfer the feature points of the previous partitions. Fig. 9 As shown, the front and rear partitions transfer shared parking space features to maintain the aligned state; for each partition, the arrangement feature points include each corner point of the parking space shape boundary, the intersection points on the vertical auxiliary line formed in the partition transfer direction (all recorded as boundary points), and any point of any parking space in the previous order partition (recorded as the original feature point), and the point where the feature point is projected onto the auxiliary line of the next partition along the partition transfer direction (recorded as the projected feature point).

[0118] Specifically, the parking spaces are arranged in two ways, horizontally and vertically, through each feature point, i.e., the parking spaces are arranged horizontally or vertically with the main direction of the partition, and the multiple partition parking space schemes are arranged as follows: Fig.10 As shown; the corner points of the parking spaces are aligned with the selected feature points for arrangement, the number of parking spaces that completely fall into the partition is calculated, and a multiplication coefficient of the corresponding proportion of the scheme arranged with the front and back feature points is given to improve the priority of the parking space alignment scheme, and finally the arrangement scheme with the largest number of parking spaces is retained.

[0119] Specifically, the number of parking spaces for the front and rear feature point solutions N = N 实际 *(1+e); where e is a variable parameter and e≥0. By adjusting e, the priority of the parking space alignment scheme can be strengthened or weakened.

[0120] In one embodiment, based on the traversal order of each node, randomly selecting a direction from the smallest column grid unit to perform expansion connection to form a partition, and determining several partition division schemes includes the following steps:

[0121] S311, based on the traversal order of each node, randomly select one-way or two-way, and expand the connection with the adjacent column grid units starting from the column grid unit with the smallest sequence number according to the selected direction to form a partition;

[0122] S312, expanding the connection again according to the smallest serial number of the column grid units remaining after each expansion and connection, until all column grid units are connected into partitions;

[0123] S313. Based on the maximum number of generated solutions, repeat steps S31 and S32 to obtain several partition division solutions.

[0124] Specifically, in the above embodiment, the partition connection method is: for common quadrilateral column grid units, there are two main directions, ⑥ according to the column grid unit traversal order generated in step S2, the selected direction is extended to the adjacent unit for connection to form a partition, and the result is as follows Figure 6 shown.

[0125] Specifically, a single or two directions are selected, and a direction is randomly selected starting from the unit with the smallest serial number and then expanded to form a partition. The smallest serial number of the remaining column grid units after each expansion is expanded again until all are connected into a partition.

[0126] Specifically, the column grid unit traversal order generated in step S2 is used to obtain random column grid unit numbers, and the column grid units are connected in combination with the breadth-first algorithm to form partitions, thereby obtaining a variety of partition division schemes, which provide a basis for subsequent generation of multiple schemes. Examples of partition division schemes are as follows: Figure 7 shown.

[0127] Specifically, since any search algorithm will have a lot of results and take a lot of time when the garage plan is complex, in order to maintain the computational feasibility of solution generation within a certain period of time, the number of solution generation can be set. In this embodiment, the number of solution generation is set to 10. When the search algorithm has obtained 10 solutions, the calculation will stop and the solution will be output and passed to the subsequent steps. In the present invention, similar methods are used to control the huge number of potential solutions to ensure the time applicability of actual engineering use.

[0128] In one embodiment, the arranged feature points include boundary points, original feature points, and projected feature points;

[0129] The boundary points include each corner point of the parking space shape boundary and the intersection points on the vertical auxiliary lines formed in the partition transfer direction;

[0130] The original feature points include any point of any parking space in the partition of the previous order;

[0131] The projected feature points include points where the current feature points are projected onto the auxiliary line of the next partition along the partition transfer direction.

[0132] In one embodiment, based on the optimal parking space arrangement plan, a lane parking space arrangement is formed, a single channel is generated using a dynamic programming algorithm, a full map lane is generated according to a backtracking algorithm, and determining all parking space types includes the following steps:

[0133] S41, based on the optimal parking space arrangement plan, forming a lane parking space arrangement and establishing a parking space map structure;

[0134] Specifically, ⑨ in the partition and parking space arrangement, the lane parking space arrangement is performed, that is, the vehicle channel is generated; for all parking spaces, a parking space graph node (Node) is created, and the connection between parking spaces with parallel movable relationships is recorded as an edge (Edge), and for different parallel movable relationships, the long edge adjacent and the short edge adjacent are set as different edge attributes (Attribute), becoming an undirected graph containing nodes, connecting edges and connection relationships.

[0135] Specifically, the parallel movable relationship means that the AGV needs to remain parallel when moving in a single direction. Therefore, the AGV can be moved in parallel only when the short sides or long sides of two parking spaces are aligned in parallel (not necessarily collinear).

[0136] Specifically, Fig.11 As shown in the figure, the parking space node connection and edge attribute setting (using 0 and 1 to represent direction information) finally obtain the whole parking space graph structure connection as shown in the figure. Fig.12 shown.

[0137] S42, based on the parking space map structure, using dynamic programming algorithm to generate a single channel;

[0138] S43, using a backtracking algorithm, backtracking the last multiple-selection node of a single lane, and selecting an unselected direction to generate a new channel;

[0139] Specifically, Through the backtracking algorithm, the last multi-selected node of a single lane is backtracked, and the unselected direction is selected to generate a new channel. This process is repeated to complete the lane generation of the entire map.

[0140] Specifically, the backtracking algorithm used in the lane of the present invention includes random selection of multiple nodes with the same score, ending the generation of a single channel after sticking to the wall, and backtracking to the previous multi-selected node to reselect the direction, such as Fig.16 As shown, the numbers in the figure represent the scores of the next parking space selection in different directions.

[0141] S44, repeat step S43 until the full map lanes are generated and all parking space types are determined, including channel parking spaces and non-channel parking spaces.

[0142] Specifically, after a lane is generated by the dynamic planning due to the end of the channel due to the wall, the backtracking algorithm can return to the adjacent previous node with multiple choices (multiple potential parking spaces have the same score) to recalculate and select the route until the next backtracking is triggered, until every non-alley parking space is adjacent to an alley parking space, that is, every parking space in the entire map is reachable. After multiple backtrackings, the channels of the entire map are as follows Fig.17 As shown in the figure, black represents the channel and white represents the parking space.

[0143] In one embodiment, based on the parking space map structure, using a dynamic programming algorithm, generating a single channel includes:

[0144] S421, starting to arrange parking spaces based on the parking space map structure, taking the parking house as a starting point;

[0145] S422, after the previous parking space is generated, calculate the potential position score of the next lane parking space, and select the parking space node with the highest score to extend the lane;

[0146] The expression for calculating the potential position score of the next channel parking space is:

[0147]

[0148] Where P represents the score of the potential position of the next channel parking space; N represents the basic score, which is equal to the number of surrounding new parking spaces activated by the potential position of the next channel parking space;

[0149] S423, continuously extending the lane according to the scores of the parking spaces that have been passed and the potential parking spaces, converting the original parking spaces into lane parking spaces, until the passage reaches the parking spaces at the edge of the parking lot or the number of potential parking spaces in each next passage is less than or equal to one, thus completing the generation of a single passage.

[0150] Specifically, ⑩ use the graph structure and dynamic programming algorithm to generate a single channel. After the parking space graph structure relationship is formed through the parking space connection relationship, the parking space arrangement begins with the parking room as the starting point; after the previous parking space is generated, the number of surrounding new parking spaces N activated by the potential position of the next channel parking space is used as the basic score, and the parking space selected in the same adjacent direction as the previous pair of parking spaces (by comparing the attributes of the front and back edges, if the same is the same direction) is given a bonus of 0.5 points. If the next position is close to the edge, 0.5 points are subtracted, forming the basic logic of the dynamic programming algorithm of the present invention. The dynamic process planning is as follows: Fig.15 As shown, the numbers in the figure represent the scores P of the next parking space selection in different directions.

[0151] Specifically, Fig.13 The score of a channel generated as shown in the figure is calculated by the above dynamic programming algorithm as 1*2 (number of newly activated parking spaces) + 0.5 (additional points in the same direction) - 0.5 (subtraction of points for being close to the edge), for a total of 2 points; Fig.14 The score of another channel generated as shown is calculated by the dynamic programming algorithm as 1*2 (the number of newly activated parking spaces) + 0 (no extra points for different directions) - 0 (no subtraction for non-edge channels), totaling 2 points.

[0152] Specifically, the parking space node with the highest score or one of the parking space nodes with the highest score is randomly selected for lane selection, so that the lane can be continuously extended according to the scores of the parking spaces that have been passed and the potential parking spaces, and the original parking spaces are converted into lane parking spaces until the passage reaches the parking spaces at the edge of the parking lot, or the number of activated parking spaces for each next potential choice is less than or equal to one, that is, the lane arrangement with the least gain can be eliminated to complete the generation of a lane.

[0153] In one embodiment, according to the parking space type and in combination with the result of geometric intersection operation, single-layer parking spaces and double-layer parking spaces are allocated, and laneway parking spaces are converted. Based on the statistical results of the number of laneway parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces, and the total number of parking spaces, the optimal solution is selected, including the following steps:

[0154] S51, with the goal of maximizing the number of parking spaces, allocating double-decker parking space attributes to all non-aisle parking spaces, and alleyway parking space attributes to all aisle parking spaces;

[0155] S52, setting the geometric boundary of the low-rise area, and performing geometric intersection operation with all generated parking spaces, if the generated parking space intersects with the low-rise area, changing the parking space attribute;

[0156] Changing the parking space attributes includes: if it was originally a double-layer parking space, it will be converted into a single-layer parking space; if it was originally an alley parking space, it will be converted into an empty space without a parking space;

[0157] S53. Count the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces, select the optimal solution based on the statistical results and export it.

[0158] Specifically, The allocation of single-layer and double-layer parking spaces and the conversion of laneway parking spaces are clarified through geometric calculations. After the calculations in steps S1 to S44 are completed, two types of parking spaces are obtained, namely, laneway parking spaces and non-lane parking spaces. First, all non-lane parking spaces are assigned double-layer parking space attributes, and all laneway parking spaces are assigned laneway parking space attributes to maximize the number of parking spaces.

[0159] Specifically, the geometric boundary of the low-rise area (the low-rise area is the area that does not meet the height requirement for double-story parking spaces) is then specified to perform geometric intersection operations with all generated parking spaces. If there is an intersection with the low-rise area, the parking space attributes are changed. If it was originally a double-story parking space, it is converted to a single-story parking space (maintaining the non-channel parking space characteristics); if it was originally a laneway parking space, it is converted to an empty space, that is, no parking space (maintaining the channel characteristics), such as Fig.19 As shown, (a) is an example diagram before the low-rise height calculation, and (b) is an example diagram after the low-rise height calculation.

[0160] Specifically, Through data statistics, feedback on the advantages and disadvantages of the layout of the plan is provided to help select the optimal plan and export the plan results (CAD and Excel file formats are used in this embodiment). Since the plan has computational randomness in the partition generation, partition order generation, parking space generation within the partition, and the final channel generation process, multiple design plans can be obtained. These plans are different in terms of single- and double-layer parking spaces, lane parking spaces, total number of parking spaces, and channel arrangements. The existing computing platform (the grasshopper platform is used in this embodiment) is used to count the number of these data and display them in the lower left corner of the visualization plan. The plan visualization and the number of different types of parking spaces are shown in the figure. Fig.18 As shown, it helps designers to judge the pros and cons of solutions, determine the best solution and derive the solution results.

[0161] In summary, with the help of the above technical solution of the present invention, by introducing advanced graph structures and intelligent algorithms, the design process is fully automated. Only by inputting basic design boundary conditions, the optimal parking space layout plan can be generated in a short time; this automated design process not only greatly shortens the design cycle, but also reduces human errors, ensuring the accuracy and reliability of the design plan; in addition, the present invention can complete a layout plan containing more than 20 partitions and more than 200 AGV parking spaces within 2 minutes. This efficient processing capability provides strong technical support for the design of large-scale AGV garages. The fast design speed means that the early preparation work of the project can be completed faster, thereby accelerating the overall progress of the project, reducing project costs, and better meeting the needs of different projects. The present invention proposes a more general and adaptable arrangement algorithm and a complete workflow, which can completely arrange the parking spaces in the AGV garage. This innovation not only improves the flexibility and applicability of the design, but also provides a guarantee for the efficient operation of the parking system and the good experience of users. The present invention can flexibly cope with various complex garage environments through graph structure and dynamic programming algorithm, and can generate the optimal parking space arrangement plan for both multi-storey garages, underground garages and open-air garages. In addition, the present invention ensures the efficient operation of the AGV by optimizing the layout of parking spaces and passages. Reasonable parking space arrangement can reduce the waiting time and driving distance of the AGV during the process of picking up and parking, thereby improving the overall efficiency of the parking system, which is particularly important for improving the user experience. Users can complete parking and picking up operations faster, enjoy more convenient and comfortable parking services, and significantly improve the design and operation level of the AGV garage.

[0162] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligently arranging parking spaces in an AGV garage based on a graph structure, characterized in that: The graph-based AGV garage parking space intelligent arrangement method comprises the following steps: S1. Structural processing of garage drawings, and extraction of geometric information layers. Based on the geometric information layers, processing of axis grid and column drawings to form column grid units; S2. construct a column grid structure based on the column grid units, traverse the adjacent column grid units in the column grid structure using a breadth-first algorithm, and assign column grid unit numbers in sequence according to the traversal order; S3. Expand and connect the column grid units according to the traversal order of the random direction to form several partition division schemes, and obtain multiple sequential partition division schemes based on the random arrangement of different orders under each partition division scheme, arrange the parking spaces in the partitions in order, and determine the optimal parking space arrangement scheme; S4. Based on the optimal parking space arrangement plan, form the lane parking space arrangement, and use the dynamic programming algorithm to generate a single channel, generate the full map lane according to the backtracking algorithm, and determine all parking space types; S5. Allocate single-layer parking spaces and double-layer parking spaces according to the parking space type and the result of geometric intersection operation, and convert the lane parking spaces. Select the optimal solution based on the statistical results of the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces.

2. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 1, characterized in that: The structured processing of the garage drawings, extracting the geometric information layer, processing the axis grid and column drawings according to the geometric information layer, and forming the column grid unit includes: S11, importing CAD drawings, separating elements in the drawings into different layers, and extracting geometric information in the layers and converting them into structured data, wherein the geometric information includes axis grid lines and column positions; S12, based on the extracted geometric information, identifying the axis grid lines in the layer and determining the axis grid intersection points; S13, based on the extracted geometric information, identifying the column positions in the layer, and corresponding them to the axis grid intersections to form the boundaries of the column grid units; S14. Select the intersection of the axis grid as the center point of the column grid unit, and divide the column grid unit into independent units based on the boundary of the column grid unit.

3. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 1, characterized in that: The column grid graph structure is an undirected graph structure, each node in the column grid graph structure represents a column grid unit, and each edge in the column grid graph structure represents a connection relationship between column grid units.

4. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 3 is characterized in that: The step of constructing a column grid structure based on column grid units, traversing adjacent column grid units in the column grid structure using a breadth-first algorithm, and assigning column grid unit serial numbers in sequence includes the following steps: S21. Based on the column grid structure, select a column grid unit whose center is closest to the center of the parking room as a starting node, and find a column grid unit that shares a common edge with the starting node to obtain a number of adjacent column grid units. S22, randomly selecting a column grid unit from adjacent column grid units as the next starting node; S23, using a breadth-first algorithm, traverse the adjacent column grid units from the starting node, and if the traversal reaches a dead end, use a backtracking algorithm to return to the previous layer of multiple-selection nodes and continue traversing; S24. Repeat steps S22 to S23, and assign column grid unit numbers in order, traverse all nodes in the graph structure layer by layer from near to far, and obtain the traversal order of each node.

5. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 1, characterized in that: The method of expanding and connecting column grid units according to the traversal order of random directions to form several partitioning schemes, and obtaining multiple sequential partitioning schemes based on the random arrangement of different orders under each partitioning scheme, arranging parking spaces in the partitions in order, and determining the optimal parking space arrangement scheme includes: S31, based on the traversal order of each node, randomly select a direction from the smallest column grid unit to expand the connection, form a partition, and determine several partition division schemes; S32. According to each partition division scheme, select the partition closest to the parking room, and randomly arrange the shared edge partitions in order to obtain several sequential partition schemes with different orders; S33. Based on the sequential partitioning schemes of different orders, the parking spaces in the partitions are arranged in order, and the maximum number of parking spaces is taken as the goal, and the arrangement feature points of the previous partition are transferred to determine the optimal parking space arrangement scheme.

6. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 5, characterized in that: The method of randomly selecting a direction for expansion and connection based on the traversal order of each node from the smallest column grid unit to form a partition and determining several partition division schemes includes the following steps: S311, based on the traversal order of each node, randomly select one-way or two-way, and expand the connection with the adjacent column grid units starting from the column grid unit with the smallest sequence number according to the selected direction to form a partition; S312, expanding the connection again according to the smallest serial number of the column grid units remaining after each expansion and connection, until all column grid units are connected into partitions; S313. Based on the maximum number of generated solutions, repeat steps S31 and S32 to obtain several partition division solutions.

7. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 5, characterized in that: The arrangement feature points include boundary points, original feature points and projected feature points; Wherein, the boundary points include each corner point of the parking space shape boundary and the intersection points on the vertical auxiliary lines formed in the partition transfer direction; The original feature point includes any point of any parking space in the partition of the previous order; The projected feature point includes a point where the current feature point is projected onto a subsequent partition auxiliary line along the partition transfer direction.

8. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 1, characterized in that: The method of forming a lane parking space arrangement based on the optimal parking space arrangement scheme, generating a single channel using a dynamic programming algorithm, generating a full-map lane according to a backtracking algorithm, and determining all parking space types includes the following steps: S41, based on the optimal parking space arrangement plan, forming a lane parking space arrangement and establishing a parking space map structure; S42, based on the parking space map structure, using dynamic programming algorithm to generate a single channel; S43, using a backtracking algorithm, backtracking the last multiple-selection node of a single lane, and selecting an unselected direction to generate a new channel; S44, repeating step S43 until the full map lanes are generated and all parking space types are determined, where the parking space types include channel parking spaces and non-channel parking spaces.

9. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 8, characterized in that: The method of generating a single channel based on the parking space map structure and using a dynamic programming algorithm includes: S421, starting to arrange parking spaces based on the parking space map structure, taking the parking house as a starting point; S422, after the previous parking space is generated, calculate the potential position score of the next lane parking space, and select the parking space node with the highest score to extend the lane; The expression for calculating the potential position score of the next channel parking space is: In the formula, P represents the score of the potential position of the next channel parking space; N represents the basic score; S423, continuously extending the lane according to the scores of the parking spaces that have been passed and the potential parking spaces, converting the original parking spaces into lane parking spaces, until the passage reaches the parking spaces at the edge of the parking lot or the number of potential parking spaces in each next passage is less than or equal to one, thus completing the generation of a single passage.

10. The method for intelligently arranging parking spaces in an AGV garage based on a graph structure according to claim 1, characterized in that: The method of allocating single-layer parking spaces and double-layer parking spaces according to the parking space type and combining the result of geometric intersection operation, and converting lane parking spaces, and selecting the optimal solution based on the statistical results of the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces includes the following steps: S51, with the goal of maximizing the number of parking spaces, allocating double-decker parking space attributes to all non-aisle parking spaces, and alleyway parking space attributes to all aisle parking spaces; S52, setting the geometric boundary of the low-rise area, and performing geometric intersection operation with all generated parking spaces, if the generated parking space intersects with the low-rise area, changing the parking space attribute; The changing of parking space attributes includes: if it is a double-layer parking space, it is converted into a single-layer parking space; if it is a lane parking space, it is converted into an empty space without parking space; S53. Count the number of lane parking spaces, the number of double-layer parking spaces, the number of single-layer parking spaces and the total number of parking spaces, select the optimal solution based on the statistical results and export it.

Images