Unmanned AGV path planning method and system based on big data

Abstract

The invention relates to the technical field of industrial control, in particular to an unmanned AGV path planning method and system based on big data, and the method comprises the steps: obtaining the position data of an AGV activity region, including a coordinate set of a starting region, a target region and a drivable region; obtaining boundary dimension parameters and motion characteristic data sets of the two types of AGVs; establishing a space-time coordinate system, cutting the position data into blocks, and generating a path layout candidate set as an initial population; iterative evolution is carried out on the initial population, the path fitness is calculated, superior individuals are selected through a roulette selection method to carry out crossover variation, and the final superior individual is used as a planning scheme. And the utilization efficiency of the specified AGV is improved by analyzing the characteristic parameters of the AGV. Through optimization of the motion efficiency and the path conflict coefficient, the operation efficiency of the specific first AGV is improved; and meanwhile, the operation time of the second AGV is dynamically adjusted through the time adjustment coefficient, so that path congestion of the first AGV caused by the second AGV is avoided.

Description

Technical Field

[0001] The present invention relates to the field of industrial control technology, and in particular to a big data-based unmanned AGV vehicle path planning method and system. Background Art

[0002] In the industrial field, unmanned AGVs are often used to transport semi-finished products between processes to reduce labor costs, and the pre-set standard procedures also improve accuracy. In actual production, AGVs often need to be customized. In essence, AGVs are a transport device equipped with specific manipulators such as forklift arms and clamp arms, etc., with the purpose of loading and unloading materials. In some occasions, AGVs do not need to load and unload goods, and only have a simple transportation function. The loading function is completed by the mechanical clamp arms equipped with the two processes it serves. However, in some occasions, AGVs must at least be equipped with mechanical actuators for lifting, bundling, loading and unloading, and the price of such AGVs is relatively high. In actual production practice, sometimes the logistics transportation between two processes requires both AGVs with specially customized manipulators and ordinary AGVs to cooperate. At present, the path planning for unmanned AGVs is for single AGVs and AGV group control. However, when the factory is equipped with a variety of AGVs, path planning is required to maximize the handling efficiency of the expensive AGV path planning. Summary of the invention

[0003] (1) Technical issues to be resolved The purpose of the present invention is to provide an unmanned AGV path planning method and system based on big data, so as to solve the path planning problem of multiple AGVs in the process of path cross-coordination and the need to give priority to the efficient operation of some AGVs.

[0004] (2) Technical solution To achieve the above objectives, on the one hand, the present invention provides a path planning method for an unmanned AGV based on big data, the method comprising: Acquire the position data of the pre-planned AGV activity area, the position data includes the starting area coordinate set, the target area coordinate set and the marked drivable area coordinate set; the AGV includes the first AGV and the second AGV, obtain the AGV's external dimension parameters, establish the motion characteristic data set of the first AGV and the second AGV, the motion characteristic data set includes the maximum speed parameter and the maximum turning radius parameter; establish a spatiotemporal coordinate system, the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; The location data is cut into blocks according to the shape dimension parameters, and an AGV path layout candidate set is generated through a spatiotemporal coordinate system, wherein the AGV path layout candidate set includes an alternative path from a starting area coordinate set to a target area coordinate set, and each alternative path is composed of a number of passable blocks connected in time sequence; the AGV path layout candidate set is used as an initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path spatiotemporal sequence, wherein the existence state takes a value of 0 or 1, and the path spatiotemporal sequence records the time point of the AGV in each passable block; The initial population is iteratively evolved, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the matching degree between the distance between adjacent passable blocks in the path time-space sequence and the corresponding AGV motion characteristic data set; selecting the winning individual by roulette selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individual that does not meet the constraint conditions of the motion characteristic data set; replacing the initial population with the offspring individual and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individual and the parent individual is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme of the AGV.

[0005] Furthermore, the method for calculating the path fitness of each individual in the initial population includes: Get the initial population Individual path planning solutions Defined as: ; in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ; in The first The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ; in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ; in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ; in is the average speed ratio: .

[0006] Furthermore, the method for obtaining the path conflict coefficient includes: Path conflict coefficient for: ; in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ; in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ; in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

[0007] Furthermore, the method for generating the path spatiotemporal sequence includes: Get the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ; in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ; in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

[0008] Furthermore, the method further comprises: Set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

[0009] Based on the same inventive concept, on the other hand, the present invention also provides an unmanned AGV vehicle path planning system based on big data, the system comprising: The position acquisition module is used to acquire the position data of the pre-planned AGV activity area, and the position data includes a starting area coordinate set, a target area coordinate set, and a marked drivable area coordinate set; the AGV includes a first AGV and a second AGV, and the AGV's outer dimension parameters are acquired to establish a motion characteristic data set of the first AGV and the second AGV, and the motion characteristic data set includes a maximum speed parameter and a maximum turning radius parameter; a spatiotemporal coordinate system is established, and the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; A parameter initialization module is used to cut the position data into blocks according to the shape dimension parameters, and generate an AGV path layout candidate set through a spatiotemporal coordinate system, wherein the AGV path layout candidate set includes an alternative path from a starting area coordinate set to a target area coordinate set, and each alternative path is composed of a number of passable blocks connected in chronological order; the AGV path layout candidate set is used as an initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path spatiotemporal sequence, wherein the existence state takes a value of 0 or 1, and the path spatiotemporal sequence records the time point of the AGV in each passable block; A path optimization module is used to iteratively evolve the initial population, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the distance between adjacent passable blocks in the path time-space sequence and the matching degree of the corresponding AGV motion characteristic data set; selecting the winning individual by roulette selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individual that does not meet the constraint conditions of the motion characteristic data set; replacing the initial population with the offspring individual and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individual and the parent individual is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme of the AGV.

[0010] Furthermore, the system further comprises: Fitness calculation module, used to obtain the first Individual path planning solutions Defined as: ; in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ; in The first The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ; in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ; in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ; in is the average speed ratio: .

[0011] Furthermore, the system further comprises: AGV conflict simulation calculation module, used to calculate the path conflict coefficient for: ; in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ; in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ; in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

[0012] Furthermore, the system further comprises: Path spatiotemporal sequence generation module, used to obtain the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ; in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ; in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

[0013] Furthermore, the system further comprises: AGV position fine-tuning module, used to set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

[0014] (3) Beneficial effects Compared with the prior art, the beneficial effect of the present invention is to improve the utilization efficiency of the specified AGV by analyzing the characteristic parameters of the AGV. The operation efficiency of the specific first AGV is improved by optimizing the motion efficiency and the path conflict coefficient; at the same time, the operation time of the second AGV is dynamically adjusted through the time adjustment coefficient to avoid the second AGV from blocking the path of the first AGV. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a method block diagram of a method for unmanned AGV vehicle path planning based on big data according to Embodiment 1 of the present invention; Figure 2 This is a module block diagram of an unmanned AGV vehicle path planning system based on big data according to Embodiment 2 of the present invention; Figure 3It is a schematic diagram of an application scenario of an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Before giving examples, it is necessary to explain the application scenarios of the present invention. Take a factory that produces engine crankshafts as an example. To transport special-shaped engine crankshafts and ensure that there is no damage between the two crankshafts during the transportation process, it is necessary to ensure that the bundling and positioning device of the AGV is perforated and positioned to the crankshaft hole. At the same time, the AGV is equipped with an anti-collision slot that separates the two crankshafts. This type of Class I customized AGV has precision requirements for the bundling, loading and unloading processes, and the cost is about 220,000 per unit. These processes need to be interspersed with Class II ordinary AGVs, such as loading crankshaft blanks from the foundry workshop to the milling workshop. The transportation of blank materials does not need to consider damage, and there is no need to customize dedicated bundling, loading and unloading robotic arms. Therefore, it is chosen to purchase more economical Class II ordinary AGVs, which cost about 25,000 each. The loading and unloading are also completed by the robotic arms equipped for the milling process. Figure 3 The schematic diagram shows that two types of AGVs cross in the work area, where the second AGV delivers the blank to the turning feed area, and the first AGV delivers the crankshaft from the turning discharge area to the milling feed area. The milling workshop of the factory is equipped with about 50 milling CNC machine tools. AGVs are required for the loading and unloading of blanks and semi-finished products of each machine tool after the processing process. The Class I AGVs and Class II AGVs corresponding to these 50 milling CNC machine tools have intersections in actual path planning. During the planning process, algorithm optimization is required to ensure that the handling efficiency of the path planning of the Class I AGV is maximized, while at the same time, a suitable number of Class II AGVs can be purchased and their tasks can be successfully completed.

[0018] Example 1: Figure 1 As shown, this embodiment provides an unmanned AGV vehicle path planning method based on big data, and the method includes: S1. Obtain the position data of the pre-planned AGV activity area, the position data includes a starting area coordinate set, a target area coordinate set and a marked drivable area coordinate set; the AGV includes a first AGV and a second AGV, obtain the AGV's external dimension parameters, establish the motion characteristic data set of the first AGV and the second AGV, the motion characteristic data set includes a maximum speed parameter and a maximum turning radius parameter; establish a spatiotemporal coordinate system, the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; S2. Cut the location data into blocks according to the shape dimension parameters, and generate an AGV path layout candidate set through the space-time coordinate system, wherein the AGV path layout candidate set includes an alternative path from the starting area coordinate set to the target area coordinate set, and each alternative path is composed of a number of passable blocks connected in time sequence; the AGV path layout candidate set is used as the initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path space-time sequence, wherein the existence state takes a value of 0 or 1, and the path space-time sequence records the time point of the AGV in each passable block; S3. Iteratively evolve the initial population, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the distance between adjacent passable blocks in the path space-time sequence and the degree of matching of the corresponding AGV motion characteristic data set; selecting the winning individual through the roulette wheel selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individuals that do not meet the constraints of the motion characteristic data set; replacing the initial population with the offspring individuals and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individuals and the parent individuals is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme for the AGV.

[0019] In order to better illustrate the present embodiment, illustratively, an AGV path planning experiment is conducted in a certain engine crankshaft production plant. The plant has 50 CNC lathes and 15 CNC milling machines. First, the workshop data is collected. The total area of ​​the factory workshop is 100 meters long and 100 meters wide. The loading station coordinates of the 50 CNC lathes are measured by a laser rangefinder as the target area coordinate set of the second AGV (the starting area coordinate set of the second AGV is the blank stacking area), and the corresponding 50 CNC lathe unloading station coordinates are measured as the starting area coordinate set of the first AGV, and the feeding station coordinates of the 15 CNC milling machines are used as the target area coordinate set of the first AGV. The total length of the magnetic strip navigation belt laid along the workshop floor covers an area of ​​about 3,200 meters (considering the exclusion of other non-working areas in the workshop), constituting a drivable area coordinate set. Two types of AGVs were used for path planning tests in the experiment: the first type was a customized AGV (referred to as the first AGV) specially used for handling fine-processed crankshafts, with measured dimensions of 3 meters long, 1.8 meters wide, and 1.6 meters high. The maximum speed was 1.2 meters per second under full load conditions, and the maximum turning radius was 2.5 meters; the second type was a common AGV (referred to as the second AGV) for handling crankshaft blanks, with measured dimensions of 2.5 meters long, 1.5 meters wide, and 1.4 meters high. The maximum speed was 1.5 meters per second under full load conditions, and the maximum turning radius was 2 meters. The origin was set at the southwest corner of the workshop to establish a rectangular coordinate system, extending 100 meters to the east in the positive direction of the x-axis and 100 meters to the north in the positive direction of the y-axis. The planning start time was selected as time zero, and the experimental sampling time interval was set to 0.5 seconds to form a three-dimensional space-time coordinate system.

[0020] According to the larger size of the first AGV (3 meters × 1.8 meters), the workshop area is divided into 625 4 meters × 4 meters squares, each of which is a passable block. These blocks are numbered 1-625 from left to right and from bottom to top. 100 candidate paths are generated through spatiotemporal coordinate measurement. Each candidate path goes from the starting area coordinate set to the target area coordinate set, and the passable block numbers passed by the AGV are recorded along the way. For example: a candidate path is recorded as a block sequence "125-126-127-152-177-178", which means that the AGV starts from block number 125, passes through blocks 126, 127, 152, and 177 in sequence, and finally reaches block 178. These 100 candidate paths constitute the initial population. Each path plan records: the AGV type (1 represents the first AGV, 2 represents the second AGV), the existence status of whether the AGV is enabled (1 represents enabled, 0 represents not enabled), and the specific time point when the AGV arrives at each block.

[0021] The initial population is iteratively evolved and optimized. First, the fitness of each path is calculated, and it is evaluated by measuring the ratio of the actual running speed between adjacent blocks to the maximum speed of the AGV, and whether the turning radius meets the limit. Among the 100 initial solutions in the first generation, the average fitness is 0.72; the roulette method is used to select solutions with higher fitness, and these solutions are crossover and mutated to generate new optimization solutions. During the optimization process, all solutions that do not meet the AGV motion characteristic restrictions (such as: the speed between adjacent blocks exceeds the maximum speed limit, the turning radius exceeds the maximum value, etc.) are eliminated. In this way, 80 optimization solutions are retained in the 5th generation, and the average fitness is increased to 0.81; 60 solutions are retained in the 10th generation, and the average fitness reaches 0.86; 40 solutions are retained in the 15th generation, and the average fitness is increased to 0.89; 20 solutions are retained in the 20th generation, and the average fitness reaches 0.91.

[0022] When the fitness improvement value is lower than the preset 0.1% threshold, the iteration stops. This experiment reached the termination condition at the 23rd generation, and finally selected the path plan with the highest fitness as the planning result. The fitness of this optimal plan is 0.93, with an average running time of 180 seconds for the first AGV and 150 seconds for the second AGV. The total path length is 22% shorter than the initial planning plan.

[0023] Furthermore, the method for calculating the path fitness of each individual in the initial population includes: Get the initial population Individual path planning solutions Defined as: ; in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ; in The first The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ; in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ; in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ; in is the average speed ratio: .

[0024] For example, in the AGV path planning experiment in the above engine crankshaft production plant, it is necessary to accurately calculate the path fitness. The 23rd generation winning individual with the highest fitness is selected for calculation example. This individual is a path planning scheme containing 12 AGVs (5 first AGVs and 7 second AGVs). The type identification and existence status of each AGV are recorded as follows: the type identification of the 5 first AGVs is 1 and the existence status is 1; the type identification of the 7 second AGVs is 2, of which the existence status of 5 is 1, and the existence status of 2 is 0 because they are not put into use yet. The path space-time sequence of each AGV records the block coordinates at different time points. For example, the first AGV starts from the No. 1 discharge station of the CNC lathe (block coordinate 125), passes through blocks 126, 127, and 152 in turn, and finally arrives at the No. 1 feed station of the CNC milling machine (block 178). The time points of arriving at each block are 0 seconds, 15 seconds, 30 seconds, 45 seconds, and 60 seconds respectively.

[0025] The calculation process of the path fitness of the individual is as follows: First, calculate the motion efficiency of each AGV between adjacent blocks. Taking the first AGV mentioned above as an example, the distance from block 125 to block 126 is 4 meters, which takes 15 seconds, and the maximum speed of this type of AGV is 1.2 meters per second, and the speed ratio is 0.222 (calculation method: 4 meters ÷ (15 seconds × 1.2 meters per second)). In this way, all speed ratios of the AGV on the entire path are calculated: the speed ratio of blocks 126 to 127 is 0.222, the speed ratio of blocks 127 to 152 is 0.185, and the speed ratio of blocks 152 to 178 is 0.198. The average value of this group of speed ratios is 0.207, and the variance is 0.0003. Substituting into the motion efficiency calculation formula, the path motion efficiency of the AGV is 0.207 / √(1+0.0003)=0.206.

[0026] In this path planning scheme, the motion efficiency of other AGVs is calculated in the same way. For example, a second AGV starts from the blank stacking area (block 33), passes through blocks 34 and 59, and arrives at the CNC lathe No. 2 loading station (block 84), with arrival times of 0 seconds, 12 seconds, 24 seconds, and 36 seconds respectively. The speed ratios of each section are 0.222, 0.222, and 0.198 respectively (calculation method: adjacent block distance ÷ (time interval × 1.5 m / s), 1.5 m / s is the maximum speed of the second AGV). The average speed ratio of the second AGV is 0.214, the variance is 0.0002, and the path motion efficiency is 0.214 / √(1+0.0002)=0.213.

[0027] Finally, the overall path fitness is calculated. The experiment shows that the minimum safe distance between two AGVs when they meet should be no less than 6 meters. Through calculation, it is found that the distance between any two AGVs in this scheme at any time is greater than 6.2 meters, so the path conflict coefficient is set to 0.1 (the preset physical space congestion coefficient is 2.0). The path motion efficiency of all enabled AGVs (10 units) is summed and divided by the number of AGVs, and then multiplied by e to the negative power of 0.2 (0.1×2.0=0.2), and the fitness of the path planning scheme is finally obtained to be 0.93.

[0028] Furthermore, the method for obtaining the path conflict coefficient includes: Path conflict coefficient for: ; in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ; in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ; in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

[0029] For example, in the AGV path planning experiment in the above-mentioned engine crankshaft production plant, it is necessary to accurately evaluate the path conflicts between different AGVs. Continuing with the 23rd generation winning individuals as an example, the path conflicts that may occur between 10 enabled AGVs (5 first AGVs and 5 second AGVs) are evaluated. First, the minimum safety distance threshold between any two AGVs is calculated: the minimum safety distance between the two first AGVs is 4 meters (considering the sensitivity of the finished crankshaft products to collisions), the minimum safety distance between a first AGV and a second AGV is 3.5 meters (considering the safety margin under mixed working conditions), and the minimum safety distance between the two second AGVs is 3 meters (considering the relatively low anti-collision requirements of the blanks).

[0030] Through calculation, it is found that there are local conflicts in the Euclidean distances between different AGVs. For example, at the 15th second, the first AGV starting from the No. 1 discharge station of the CNC lathe is located in block 126, while the second AGV starting from the blank stacking area is located in block 34. The Euclidean distance between the two is 3.2 meters, which is less than the 3.5-meter minimum safety distance threshold specified between them. The spatiotemporal conflict evaluation coefficient is recorded as 1. A total of 12 local conflict points were found in this path planning scheme. These conflict points mainly occurred in the narrow passage area in the middle of the workshop, involving 3 first AGVs and 4 second AGVs. After summing the spatiotemporal conflict evaluation coefficients of these 12 conflicts and dividing them by the number of all possible path point combinations (assuming that each path consists of 50 points, the total number of combinations is 49×49=2401), the path conflict coefficient of this scheme is 0.005.

[0031] In order to reduce path conflicts, the experiment evaluates the turning constraints of each sub-path. Taking the first AGV from the CNC lathe No. 1 discharge station to the CNC milling machine No. 1 feed station as an example, its path sequence "125-126-127-152-177-178" contains two turns: blocks 127 to 152 are 90 degrees northward, and blocks 152 to 177 are 90 degrees eastward. The turning radius obtained by calculating the three consecutive points "127-152-177" is 2.3 meters, which is less than the maximum turning radius of 2.5 meters allowed for this type of AGV. The same calculation is performed on other turning points, and the results show that all turning radii are within the safe range, and the turning constraint degree of this path is 0.95.

[0032] Finally, the turning constraint of the 10 sub-paths is multiplied by the existence status of each AGV (all 1), and the path turning constraint is 9.5. This value shows that the path planning scheme can basically avoid the instability of AGV operation caused by sharp turns while satisfying the turning radius constraint. The experimental results show that by setting the turning constraint reasonably, the risk of path conflict between different AGVs can be effectively reduced.

[0033] Furthermore, the method for generating the path spatiotemporal sequence includes: Get the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ; in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ; in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

[0034] For example, in the AGV path planning experiment of the above-mentioned engine crankshaft production plant, it is necessary to determine the specific travel time points of each AGV in the path space-time sequence. Continuing to use the 23rd generation winning individuals as the analysis object, first calculate the total path travel time. Take the path of the first AGV from the No. 1 discharge station of the CNC lathe to the No. 1 feed station of the CNC milling machine as an example. The path passes through the block sequence "125-126-127-152-177-178", and the distances between adjacent blocks are 4 meters, 4 meters, 5 meters, and 4 meters respectively. According to the maximum speed of the first AGV of 1.2 meters per second, the ideal travel time is 14.17 seconds. Similarly, the ideal travel time of other AGVs is calculated. For example, the path of the second AGV from the blank stacking area to the No. 2 loading station of the CNC lathe is "33-34-59-84". The distances between adjacent blocks are 4 meters, 5 meters, and 5 meters. Calculated at the maximum speed of 1.5 meters per second, the ideal travel time is 9.33 seconds.

[0035] When determining the specific time point at which each AGV arrives at each block, first set the starting time to 0 seconds. For the first AGV, the distance from block 125 (starting point) to block 126 is 4 meters. According to the maximum speed of 1.2 meters per second and the initial time adjustment coefficient of 0, the arrival time is 3.33 seconds. However, when calculating the time to reach block 127, it was found that there was a second AGV in the adjacent area at this time, and the Euclidean distance between the two was 3.2 meters, which was less than the minimum safety distance threshold of 3.5 meters. Since the maximum speed of the second AGV (1.5 meters per second) is greater than that of the first AGV (1.2 meters per second), the time adjustment coefficient of 0.2 is added to the first AGV, extending its arrival time at block 127 to 7.67 seconds.

[0036] The same method is used to calculate the arrival time of subsequent blocks. When the AGV runs to block 152, it is found that the Euclidean distance with another first AGV is 3.8 meters. At this time, there is no need to adjust the time (because it is greater than the minimum safety distance of 4 meters). Finally, the complete path time-space sequence of the first AGV is: block 125 (0 seconds), block 126 (3.33 seconds), block 127 (7.67 seconds), block 152 (11.83 seconds), block 177 (15.17 seconds), block 178 (18.50 seconds). In this way, a specific time schedule is formulated for each AGV, which effectively avoids mutual interference during operation. Experimental data show that after considering the time adjustment coefficient, the actual running time of the AGV increases by an average of 15%, but the number of path conflicts decreases by 85%.

[0037] Furthermore, the method further comprises: Set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value and corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value and corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

[0038] For example, in the AGV path planning experiment of the above-mentioned engine crankshaft production plant, the path optimization is performed on the winning individuals of the 23rd generation. The workshop area is set as a two-dimensional coordinate plane, and the actual running coordinates corresponding to the path space-time sequence "125-126-127-152-177-178" of the first AGV are measured as follows: (20,25), (24,25), (28,25), (28,38), (35,38), (35,42), in meters. Considering the dimensions of the first AGV (3 meters long × 1.8 meters wide), the offset threshold in the x-axis direction is set to 2 meters (that is, the minimum distance between the center of the AGV and the edge of the workshop), and the offset threshold in the y-axis direction is also set to 2 meters. The offset threshold in the x-axis direction is divided into 20 equal parts, and the first offset step length is 0.1 meters; the offset threshold in the y-axis direction is divided into 20 equal parts, and the second offset step length is 0.1 meters.

[0039] First, the path of the first AGV is optimized. Along the x-axis direction, the first coordinate point (20,25) on the path is moved to the right by 0.1 meters, and the calculated motion efficiency is 0.206; continue to move to the right by 0.1 meters, and the motion efficiency rises to 0.208; repeat this process, and when it moves to (20.4,25), the motion efficiency reaches the maximum value of 0.212. Using the same method, the other coordinate points on the path are optimized in the x-axis direction, and the cumulative first offset steps of the x-axis are: 0.4 meters, 0.3 meters, 0.2 meters, 0.3 meters, and 0.4 meters. Then, similar optimization is performed in the y-axis direction, and the first coordinate point (20.4,25) on the path is moved upward by 0.1 meters. The optimal position of each coordinate point in the y-axis direction is obtained by the gradient descent method, and the cumulative second offset steps of the y-axis are: 0.3 meters, 0.2 meters, 0.3 meters, 0.2 meters, and 0.3 meters. The optimized running coordinates of the first AGV are corrected to: (20.4, 25.3), (24.3, 25.2), (28.2, 25.3), (28.3, 38.2), (35.3, 38.2), (35.4, 42.3).

[0040] After completing the path optimization of the first AGV, the second AGV is optimized. Taking the second AGV from the blank stacking area to the No. 2 loading station of the CNC lathe as an example, its original path coordinates are: (6,8), (6,12), (12,12), (12,16). Using the same optimization method as the first AGV, while ensuring that the safety distance with the optimized first AGV path is not less than 3.5 meters, the optimized coordinates are: (6.2,8.3), (6.3,12.2), (12.2,12.3), (12.3,16.2). Through path optimization, the motion efficiency of the first AGV is increased from 0.206 to 0.212, and the motion efficiency of the second AGV is increased from 0.213 to 0.218, while ensuring that the safety distance between different AGVs always meets the requirements.

[0041] Embodiment 2: Based on the same inventive concept, Figure 2 As shown, this embodiment also provides an unmanned AGV vehicle path planning system based on big data, and the system includes: The position acquisition module is used to acquire the position data of the pre-planned AGV activity area, and the position data includes a starting area coordinate set, a target area coordinate set, and a marked drivable area coordinate set; the AGV includes a first AGV and a second AGV, and the AGV's outer dimension parameters are acquired to establish a motion characteristic data set of the first AGV and the second AGV, and the motion characteristic data set includes a maximum speed parameter and a maximum turning radius parameter; a spatiotemporal coordinate system is established, and the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; A parameter initialization module is used to cut the position data into blocks according to the shape dimension parameters, and generate an AGV path layout candidate set through a spatiotemporal coordinate system, wherein the AGV path layout candidate set includes an alternative path from a starting area coordinate set to a target area coordinate set, and each alternative path is composed of a number of passable blocks connected in chronological order; the AGV path layout candidate set is used as an initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path spatiotemporal sequence, wherein the existence state takes a value of 0 or 1, and the path spatiotemporal sequence records the time point of the AGV in each passable block; A path optimization module is used to iteratively evolve the initial population, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the distance between adjacent passable blocks in the path time-space sequence and the matching degree of the corresponding AGV motion characteristic data set; selecting the winning individual by roulette selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individual that does not meet the constraint conditions of the motion characteristic data set; replacing the initial population with the offspring individual and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individual and the parent individual is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme of the AGV.

[0042] Furthermore, the system further comprises: Fitness calculation module, used to obtain the first Individual path planning solutions Defined as: ; in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ; in The first one is located in the AGV activity area. The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ; in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ; in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ; in is the average speed ratio: .

[0043] Furthermore, the system further comprises: AGV conflict simulation calculation module, used to calculate the path conflict coefficient for: ; in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ; in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ; in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

[0044] Furthermore, the system further comprises: Path spatiotemporal sequence generation module, used to obtain the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ; in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ; in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

[0045] Furthermore, the system further comprises: AGV position fine-tuning module, used to set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value and corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value and corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

[0046] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0047] Finally, it should be noted that: Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A path planning method for unmanned AGV based on big data, characterized in that: The method comprises: Acquire the position data of the pre-planned AGV activity area, the position data includes the starting area coordinate set, the target area coordinate set and the marked drivable area coordinate set; the AGV includes the first AGV and the second AGV, obtain the AGV's external dimension parameters, establish the motion characteristic data set of the first AGV and the second AGV, the motion characteristic data set includes the maximum speed parameter and the maximum turning radius parameter; establish a spatiotemporal coordinate system, the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; The location data is cut into blocks according to the shape dimension parameters, and an AGV path layout candidate set is generated through a spatiotemporal coordinate system, wherein the AGV path layout candidate set includes an alternative path from a starting area coordinate set to a target area coordinate set, and each alternative path is composed of a number of passable blocks connected in time sequence; the AGV path layout candidate set is used as an initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path spatiotemporal sequence, wherein the existence state takes a value of 0 or 1, and the path spatiotemporal sequence records the time point of the AGV in each passable block; The initial population is iteratively evolved, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the matching degree between the distance between adjacent passable blocks in the path time-space sequence and the corresponding AGV motion characteristic data set; selecting the winning individual by roulette selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individual that does not meet the constraint conditions of the motion characteristic data set; replacing the initial population with the offspring individual and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individual and the parent individual is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme of the AGV.

2. According to the big data-based unmanned AGV path planning method of claim 1, it is characterized in that: The method for calculating the path fitness of each individual in the initial population includes: Get the initial population Individual path planning solutions Defined as: ； in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ； in The first The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ； in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ； in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ； in is the average speed ratio: 。 3. The unmanned AGV path planning method based on big data according to claim 2 is characterized in that: The method for obtaining the path conflict coefficient includes: Path conflict coefficient for: ； in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ； in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ； in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

4. The unmanned AGV path planning method based on big data according to claim 3 is characterized in that: The method for generating the path spatiotemporal sequence includes: Get the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ； in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ； in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

5. The unmanned AGV path planning method based on big data according to claim 4 is characterized in that: The method further comprises: Set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

6. An unmanned AGV vehicle path planning system based on big data, characterized in that: The system comprises: The position acquisition module is used to acquire the position data of the pre-planned AGV activity area, and the position data includes a starting area coordinate set, a target area coordinate set, and a marked drivable area coordinate set; the AGV includes a first AGV and a second AGV, and the AGV's outer dimension parameters are acquired to establish a motion characteristic data set of the first AGV and the second AGV, and the motion characteristic data set includes a maximum speed parameter and a maximum turning radius parameter; a spatiotemporal coordinate system is established, and the spatiotemporal coordinate system The axis corresponds to the position data of the AGV activity area, the time-space coordinate system The axis corresponds to the time point at which the AGV appears at different position data; A parameter initialization module is used to cut the position data into blocks according to the shape dimension parameters, and generate an AGV path layout candidate set through a spatiotemporal coordinate system, wherein the AGV path layout candidate set includes an alternative path from a starting area coordinate set to a target area coordinate set, and each alternative path is composed of a number of passable blocks connected in chronological order; the AGV path layout candidate set is used as an initial population, and each individual in the initial population includes an AGV type identifier and an AGV path state, wherein the AGV type identifier is used to mark the first AGV or the second AGV, and the AGV path state includes an existence state and a path spatiotemporal sequence, wherein the existence state takes a value of 0 or 1, and the path spatiotemporal sequence records the time point of the AGV in each passable block; A path optimization module is used to iteratively evolve the initial population, and the iterative evolution step includes: calculating the path fitness of each individual in the initial population, the path fitness is calculated according to the distance between adjacent passable blocks in the path time-space sequence and the matching degree of the corresponding AGV motion characteristic data set; selecting the winning individual by roulette selection method according to the path fitness ranking, crossover and mutation of the winning individual to obtain the offspring individual, and eliminating the offspring individual that does not meet the constraint conditions of the motion characteristic data set; replacing the initial population with the offspring individual and repeating the iterative evolution step until the improvement value of the path fitness of the offspring individual and the parent individual is lower than the preset fitness threshold; and using the final winning individual after completing the iterative evolution as the path planning scheme of the AGV.

7. The unmanned AGV path planning system based on big data according to claim 6 is characterized in that: The system further comprises: Fitness calculation module, used to obtain the first Individual path planning solutions Defined as: ； in Representative Path planning solutions No. Type identification of each AGV, The value 1 indicates the first AGV. A value of 2 indicates the second AGV; Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions No. The spatiotemporal sequence of the paths of the AGVs, is the number of AGVs; Path space-time sequence Defined as: ； in The first The center coordinates of the traversable blocks; The time when the AGV arrives at the block; is the number of traversable blocks contained in the path space-time sequence; Calculate path fitness for: ； in For path The efficiency of movement, is the path conflict coefficient; the path conflict coefficient It is obtained by calculating the Euclidean distance of each point in the path space-time sequence of the two AGVs, and adding the calculated Euclidean distance to the minimum safety distance threshold between the two AGVs; is the preset physical space crowding coefficient; Path motion efficiency The calculation method is: ； in For the Path planning solutions No. The AGV is the first in its path space-time sequence The traversable blocks and The Euclidean distance between the center points of traversable blocks; is the maximum speed parameter of the corresponding AGV type; The difference between adjacent time points ; is the variance of the speed ratio series; the variance of the speed ratio series The calculation method is: ； in is the average speed ratio: 。 8. The unmanned AGV path planning system based on big data according to claim 7 is characterized in that: The system further comprises: AGV conflict simulation calculation module, used to calculate the path conflict coefficient for: ； in Representative Path planning solutions No. The existence status of an AGV, the value is 0 or 1; Representative Path planning solutions The The AGV is on its path Point and The AGV is on its path The spatial-temporal conflict evaluation coefficient between points: ； in For the AGVs at waypoints With AGVs at waypoints The Euclidean distance between For the The first AGV and the The minimum safety distance threshold between AGVs; There are two types of AGVs and get; Path turning constraint The calculation method is: ； in is three consecutive points in the space-time sequence of the passing path , , Calculated turning radius; It is the maximum turning radius parameter of the corresponding AGV type.

9. The unmanned AGV path planning system based on big data according to claim 8 is characterized in that: The system further comprises: Path spatiotemporal sequence generation module, used to obtain the passable block sequence of the AGV activity area , No. Path planning solutions The AGV calculates its travel time : ； in Representative Path planning solutions Middle The ideal travel time for an AGV to complete the entire path; Determine the time points in the path space-time sequence based on the travel time ,when hour, ;when hour: ； in The time adjustment coefficient with an initial value of 0 is used to dynamically adjust the Path planning solutions Middle The running time of an AGV; when the path conflict coefficient When it is detected that the Euclidean distance between two AGVs is less than the minimum safety distance threshold, the AGV with the smaller maximum speed parameter is given priority to increase its corresponding time adjustment coefficient. , extend the time interval for the AGV to reach the subsequent path points until the path conflict coefficient The obtained Euclidean distance between the two AGVs is greater than or equal to the minimum safety distance threshold.

10. The unmanned AGV path planning system based on big data according to claim 9 is characterized in that: The system further comprises: AGV position fine-tuning module, used to set the AGV activity area to Two-dimensional coordinate plane, Path planning solutions Existence status The AGV with a value of 1 records its path time-space sequence The running coordinates in The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; the running coordinates are The offset threshold in the axis direction is set to The distance between the center of the AGV's outer dimensions in the axial direction and the edge of the AGV's active area; Axis direction and The offset threshold in the axial direction is divided into a number of equally divided distances, which are recorded as the first offset step and the second offset step; The first Path planning solutions The first AGV in The first offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the first offset step; Path planning solutions The first AGV in The second offset step of the axis movement is obtained by gradient descent method Axis path motion efficiency The extreme value of and the corresponding The axis accumulates the second offset step; updates the path time-space sequence of the first AGV according to the corresponding accumulated first offset step and second offset step The position coordinates in ; After completing the path optimization of the first AGV, the path of the second AGV is optimized according to the same path optimization method as the first AGV; when the second AGV is performing path optimization, the Euclidean distance between the second AGV path time-space sequence and the optimized first AGV path time-space sequence is greater than or equal to the minimum safety distance threshold between the two AGVs.

Images