Omnidirectional AGV path planning method, device and equipment and storage medium

Abstract

The invention relates to the technical field of AGV path planning, and discloses an omnidirectional AGV path planning method, device and equipment and a storage medium, and the method comprises the steps: obtaining a dynamic environment state matrix and environment change rate data in a stereo garage; performing instantaneous queue length calculation on the task arrival rate and AGV processing capacity data in the operation period of the stereo garage to obtain an instantaneous task queue parameter and queue dissipation prediction time; performing composite trigger condition judgment on the AGV position deviation and the environment variable to obtain a dynamic event trigger signal and a task priority parameter; the dynamic event trigger signals and the task priority parameters are input into the instantaneous task allocation model to be solved, an optimal task allocation scheme and a corresponding target path set are obtained, task queues formed in a short time in the stereo garage operation process can be accurately recognized and quantified, and the robustness and adaptability of path planning are improved.

Description

Technical Field

[0001] This application relates to the technical field of AGV path planning, and particularly to a path planning method, device, equipment and storage medium for an omnidirectional AGV. Background Art

[0002] As an important means to solve the problem of parking difficulty, the stereoscopic garage has been widely used, and the omnidirectional AGV has become a key device for car handling in the stereoscopic garage due to its flexible motion characteristics. However, the traditional AGV path planning methods are mainly designed for static or quasi-static environments, and use the method of pre-calculating fixed paths for navigation. In the face of the complex and changeable dynamic environment of the stereoscopic garage, it is often unable to respond to environmental changes in a timely manner, resulting in lagging path planning and low task execution efficiency, and it is difficult to meet the high-frequency and high-density car handling requirements of modern stereoscopic garages.

[0003] The internal environment of the stereoscopic garage has a high degree of dynamics, including factors such as mutual interference caused by multiple AGVs operating simultaneously, real-time changes in the parking space state, the emergence of sudden obstacles, and the random arrival of handling tasks. These dynamic changes often form an instantaneous task queue in a short time, resulting in a significant queuing delay effect. Most of the existing path planning methods are based on static optimization of the average queue length, ignoring the important impact of the instantaneous queuing effect on the system performance, and unable to accurately predict and process these task queues that form and dissipate in a short time, resulting in a large deviation between the path planning decision and the actual operation environment. Summary of the Invention

[0004] This application provides a path planning method, device, equipment and storage medium for an omnidirectional AGV. This application can accurately identify and quantify the task queue formed in a short time during the operation of the stereoscopic garage, and improve the robustness and adaptability of path planning.

[0005] In the first aspect of this application, a path planning method for an omnidirectional AGV is provided. The path planning method for the omnidirectional AGV includes: Obtain the dynamic environment state matrix and environmental change rate data in the stereoscopic garage; Calculate the instantaneous queue length of the task arrival rate and AGV processing capacity data during the operation cycle of the stereoscopic garage according to the dynamic environment state matrix and the environmental change rate data, and obtain the instantaneous task queue parameters and the queue dissipation prediction time; Based on the instantaneous task queue parameters and the queue dissipation prediction time, judge the composite trigger conditions for the AGV position deviation and environmental variables, and obtain the dynamic event trigger signal and the task priority parameters; Input the dynamic event trigger signal and the task priority parameters into the instantaneous task allocation model for solution, and obtain the optimal task allocation plan and the corresponding target path set.

[0006] In combination with the first aspect, in the first implementation manner of the first aspect of the present invention, the obtaining of the dynamic environment state matrix and the environmental change rate data in the stereoscopic garage includes: Performing real-time data collection on lidar sensors, ultrasonic sensors, and vision sensors for each parking space layer in the stereoscopic garage to obtain obstacle position information, real-time AGV position information, parking space occupancy status information, and passageway passability information; Constructing a dynamic environment state matrix based on the obstacle position information, the real-time AGV position information, the parking space occupancy status information, and the passageway passability information; Performing state difference calculation within a continuous time window based on the dynamic environment state matrix to obtain environmental change rate data.

[0007] In combination with the first aspect, in the second implementation manner of the first aspect of the present invention, the calculating of the instantaneous queue length for the task arrival rate and the AGV processing capacity data during the operation cycle of the stereoscopic garage based on the dynamic environment state matrix and the environmental change rate data to obtain the instantaneous task queue parameters and the queue dissipation prediction time includes: Statistically analyzing the handling task flow during the operation cycle of the stereoscopic garage based on the dynamic environment state matrix and the environmental change rate data to obtain the task arrival rate and the AGV processing capacity data; Calculating the maximum value of the instantaneous queue length according to the task arrival rate and the AGV processing capacity data to obtain the instantaneous task queue length data and the queue delay weight coefficient; Calculating the task demand fluctuation intensity based on the instantaneous task queue length data and the queue delay weight coefficient to generate instantaneous task queue parameters; Performing queue dissipation time prediction calculation based on the instantaneous task queue parameters to obtain the queue dissipation prediction time.

[0008] In combination with the first aspect, in the third implementation manner of the first aspect of the present invention, the calculating of the task demand fluctuation intensity based on the instantaneous task queue length data and the queue delay weight coefficient to generate instantaneous task queue parameters includes: Performing instantaneous queuing effect quantification calculation on the instantaneous task queue length data and the queue delay weight coefficient to obtain queuing effect quantification data; Performing historical task arrival rate sequence analysis based on the queuing effect quantification data to obtain the task arrival rate average value and the standard deviation data; Performing task demand fluctuation intensity variance calculation based on the task arrival rate average value and the standard deviation data to obtain the task demand fluctuation intensity data and the fluctuation threshold comparison result; Determine the instantaneous queuing effect based on the task requirement fluctuation intensity data and the comparison result of the fluctuation threshold, and generate instantaneous task queue parameters.

[0009] Combined with the first aspect, in the fourth implementation manner of the first aspect of the present invention, the compound trigger condition judgment of the AGV position deviation and the environmental variable based on the instantaneous task queue parameter and the queue dissipation prediction time to obtain a dynamic event trigger signal and a task priority parameter includes: Calculate the Euclidean distance between the actual position and the planned position of the AGV to obtain the AGV position deviation, and perform weighted summation of the environmental variables based on the instantaneous task queue parameter to obtain the environmental variable; Perform adaptive threshold dynamic adjustment according to the ratio of the queue dissipation prediction time to the operation cycle time to obtain a trigger threshold adjustment coefficient; Perform a compound trigger condition comparison and judgment based on the AGV position deviation, the environmental variable, and the trigger threshold adjustment coefficient to obtain a dynamic event trigger signal; Evaluate the priority of the AGV position deviation, the environmental variable, and the task urgency data to obtain a task priority parameter.

[0010] Combined with the first aspect, in the fifth implementation manner of the first aspect of the present invention, inputting the dynamic event trigger signal and the task priority parameter into the instantaneous task allocation model for solution to obtain an optimal task allocation scheme and a corresponding target path set includes: Construct a first fixed-point problem of garage road network loading based on the dynamic event trigger signal and solve the AGV traffic distribution to obtain a traffic distribution vector, construct a second fixed-point problem of garage road network loading based on the task priority parameter and solve the task allocation vector to obtain a task allocation vector; Calculate the path selection probability of the traffic distribution vector and the task allocation vector through the instantaneous task allocation model to obtain task selection probability distribution data; Input the task selection probability distribution data into a variational inequality solver for cost mapping calculation, and perform path decoding and task allocation mapping calculation under the constraint conditions of the feasible solution set to obtain an optimal task allocation scheme and a corresponding target path set.

[0011] Combined with the first aspect, in the sixth implementation manner of the first aspect of the present invention, calculating the path selection probability of the traffic distribution vector and the task allocation vector through the instantaneous task allocation model to obtain task selection probability distribution data includes: Input the flow distribution vector and the task assignment vector into the instantaneous task assignment model, and calculate the path utility function values of each candidate path through the path passing time, instantaneous queuing delay, and task completion benefit to obtain path utility data; Determine the probability distribution adjustment coefficient according to the decision-making uncertainty degree of the AGV handling task in the stereo garage; Perform exponential transformation calculation based on the path utility data and the probability distribution adjustment coefficient to obtain the exponential utility values of each candidate path; Perform normalized probability operation on the exponential utility values of each candidate path to obtain task selection probability distribution data.

[0012] The second aspect of the present application provides a path planning device for an omnidirectional AGV. The path planning device for the omnidirectional AGV includes: An acquisition module, configured to acquire the dynamic environment state matrix and the environmental change rate data in the stereo garage; A calculation module, configured to calculate the instantaneous queue length of the task arrival rate and the AGV processing capacity data during the operation cycle of the stereo garage according to the dynamic environment state matrix and the environmental change rate data to obtain instantaneous task queue parameters and queue dissipation prediction time; A judgment module, configured to perform a composite trigger condition judgment on the AGV position deviation and environmental variables based on the instantaneous task queue parameters and the queue dissipation prediction time to obtain a dynamic event trigger signal and task priority parameters; A solution module, configured to input the dynamic event trigger signal and the task priority parameters into the instantaneous task assignment model for solution to obtain an optimal task assignment plan and the corresponding target path set.

[0013] The third aspect of the present application provides an electronic device, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory to enable the electronic device to execute the above-mentioned path planning method for an omnidirectional AGV.

[0014] The fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored, and when it runs on a computer, it enables the computer to execute the above-mentioned path planning method for an omnidirectional AGV.

[0015] Compared with the prior art, the present application has the following beneficial effects: By constructing a dynamic environmental state matrix and calculating environmental change rate data, it is possible to accurately perceive in real time the changes in multi-dimensional environmental information such as the positions of obstacles, the states of AGVs, and the occupancy of parking spaces in a three-dimensional garage. Compared with traditional static map methods, the adaptability to complex dynamic environments is significantly improved. By establishing an instantaneous queue length calculation mechanism and a queue dissipation prediction algorithm, it is possible to accurately identify and quantify the task queues formed within a short period during the operation of a three-dimensional garage, overcoming the limitations of the prior art that only considers the average queue, and achieving precise modeling and processing of the instantaneous queuing effect. By designing a composite trigger condition judgment mechanism based on the position deviation of the AGV and environmental variables, path replanning is only initiated when necessary, avoiding the continuous calculation mode of traditional methods, and significantly reducing the consumption of computing resources while ensuring the timeliness of system response. By adopting a variational inequality solution framework combined with an instantaneous task allocation model, it is possible to simultaneously consider multiple decision dimensions such as AGV traffic distribution, task allocation, and path selection, and achieve global coordinated optimization of the multi-AGV system in a three-dimensional garage. By dynamically adjusting the trigger threshold according to the queue dissipation prediction time and calculating the probability distribution based on the path utility function, the system can adaptively adjust key parameters according to the actual operation conditions, improving the robustness and adaptability of path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0017] The structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those who are familiar with this technology to understand and read, and are not used to limit the limited conditions under which the present invention can be implemented. Therefore, they do not have any technical essence. Any modification of the structure, change of the proportional relationship, or adjustment of the size, without affecting the effects that the present invention can produce and the purposes that can be achieved, should still fall within the scope that can be covered by the technical content disclosed by the present invention.

[0018] Figure 1 is a schematic flowchart of the path planning method for an omnidirectional AGV provided by an embodiment of the present invention; Figure 2 is a schematic block diagram of the structure of the path planning device for an omnidirectional AGV provided by an embodiment of the present invention; Figure 3 is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] The flowcharts shown in the accompanying drawings are only illustrative examples, and do not necessarily include all the contents and operations / steps, nor do they necessarily need to be executed in the described order. For example, some operations / steps can also be decomposed, combined, or partially merged. Therefore, the actual execution order may change according to the actual situation.

[0021] It should also be understood that the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0022] It should be further understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Please refer to Figure 1 , an embodiment of the path planning method for an omnidirectional AGV in the embodiments of this application includes: Step 100, obtaining a dynamic environment state matrix and environmental change rate data in a three-dimensional garage; It can be understood that the execution subject of this application can be a path planning device for an omnidirectional AGV, or a terminal or a server. Specifically, it is not limited here. The embodiments of this application will be described by taking the server as the execution subject as an example.

[0023] Specifically, a multi-source heterogeneous perception system is constructed. Based on the multi-layer structural characteristics of the stereoscopic garage, lidar sensors, ultrasonic sensors, and vision sensors are deployed at key positions of each parking space and passageway to achieve high-frequency and low-latency perception data collection of key elements in the environment. The lidar captures the spatial distribution and boundary contours of obstacles in the garage in real time through its precise ranging ability. The ultrasonic sensor is used to assist in detecting the presence of obstacles at close range and at the endpoints of the passageway. The vision sensor can identify whether a parking space is occupied, the position and attitude changes of the AGV in different lanes, and enhances the understanding of the structured scene with the help of image segmentation and object detection algorithms. The original sensor data collected is preprocessed by the edge computing unit to eliminate noise and data drift, and is uniformly converted to the standard space coordinate system for cross-sensor fusion. On this basis, through the algorithm module, four types of information, namely the position of obstacles, the real-time position of the AGV, the occupancy status of parking spaces, and the passability of the passageway, are mapped into a state vector of the standard structure, and the environmental state matrix at the current moment is constructed. This matrix describes the passability and operation status of the entire stereoscopic garage in the current time slice. To capture the dynamic change characteristics of the environment, the above state matrix is recorded in a sliding time window, and the difference between the state matrices at two consecutive moments is calculated to form the change rate data. The difference operation adopts the element-wise norm calculation strategy to quantify the change amplitude of each state monitoring point. If there are some components in the change rate data that are significantly larger than other regions, it indicates that a structural dynamic change has occurred in this region, such as the addition of obstacles, the switching of the parking space state, or the deviation of the AGV trajectory. This information is used for the pre-judgment of subsequent path adjustment and task reallocation. The environmental state matrix and its corresponding change rate data together constitute the core perception data input of the input layer of the path planning system.

[0024] Step 200: Calculate the instantaneous queue length of the task arrival rate and the AGV processing capacity data during the operation cycle of the stereoscopic garage according to the dynamic environmental state matrix and the environmental change rate data, and obtain the instantaneous task queue parameters and the queue dissipation prediction time; Specifically, based on the dynamic environmental state matrix and the environmental change rate data, continuously count the handling task flow in each area during the operation cycle of the three-dimensional garage. Combine the real-time positions of AGVs in each time window with their historical records of completing handling tasks to identify the task request frequencies and execution capabilities of different channels and operation areas in each time period, so as to calculate the time distribution characteristics of the task arrival rate and the AGV processing capacity respectively. On this basis, estimate the maximum queue length according to the instantaneous difference between the task arrival rate and the AGV processing capacity, and extract the queue delay weight coefficient in combination with the historical task backlog situation. This weight reflects the impact degree of system delay on the overall operation efficiency under different queue lengths. Quantitatively evaluate the intensity of task demand fluctuations based on the statistically obtained instantaneous queue length data and the corresponding queue delay weight coefficients. Analyze the severe fluctuations of the task flow in the time dimension through the sliding window mechanism to generate instantaneous task queue parameters reflecting the current load state and processing stability of the system, and dynamically depict the risk level of possible task congestion at different time points of the system. Combine core indicators such as the current queue length, historical environmental state change trend, and task demand fluctuation intensity to construct a queue dissipation time prediction model, simulate the time period for the natural dissipation of the task queue under the current system operation state, and output the queue dissipation prediction time accordingly, which is used to judge whether the current task backlog is a short-term sudden occurrence or will exist for a long time.

[0025] On the basis of the existing environmental state analysis, jointly calculate the instantaneous task queue length data and the queue delay weight coefficient to achieve a quantitative expression of the queuing effect. By constructing a queuing effect quantification model, weighted fusion of the instantaneous task queue lengths in each time period and their corresponding delay impact degrees is carried out to obtain the queuing effect quantification data reflecting the severity of task congestion. This data reflects the scale of the current task backlog and also reflects its potential impact on the overall operation efficiency. Based on this queuing effect quantification data, conduct statistical analysis on the historical task arrival rate sequence. By tracing back the task flow fluctuation trend within a certain period of time, extract the average value and standard deviation of this sequence, which are used to characterize the task generation mode and fluctuation amplitude of the garage task system in a typical operation cycle. Calculate the variance of the task demand fluctuation intensity based on the average value and standard deviation data of the task arrival rate to quantify the deviation degree of the current task arrival mode from the historical fluctuation range, and form the task demand fluctuation intensity data representing the current fluctuation severity. Compare this task demand fluctuation intensity data with the preset fluctuation threshold, and identify whether there is a sudden and large-scale task load change currently by judging whether it exceeds the threshold, thus completing the determination process of the instantaneous queuing effect. If the current fluctuation intensity significantly exceeds the threshold, mark the current state as a high-risk or task congestion state requiring intervention, and generate instantaneous task queue parameters containing comprehensive information such as queue length, delay impact, and fluctuation degree.

[0026] Step 300: Based on the instantaneous task queue parameters and the queue dissipation prediction time, judge the composite trigger conditions for the AGV position deviation and environmental variables to obtain a dynamic event trigger signal and task priority parameters; It should be noted that a composite event perception mechanism integrating spatial deviation and system load status is constructed. In the specific implementation process, continuously obtain the actual running trajectory of the AGV and compare it with the corresponding path planning scheme in real time. By calculating the Euclidean distance between the actual position and the planned position of the AGV, obtain the spatial position deviation value of the AGV at the current moment. This deviation value reflects whether the AGV has a path deviation or delay due to sudden obstacles, channel blockages, or task overlaps. At the same time, use the queuing effect data, task fluctuation intensity index, and queue congestion level information in the instantaneous task queue parameters as weight factors to perform a weighted sum of the variables in the environment that affect the execution stability of the AGV, forming an environmental variable index that comprehensively describes the dynamic complexity of the current operation area. This environmental variable not only reflects the current task density but also reflects the potential passage risks in the area where the AGV is located. To achieve adaptive adjustment of the trigger conditions under different load states, based on the ratio relationship between the currently predicted queue dissipation time and the standard operation cycle time, dynamically calculate a trigger threshold adjustment coefficient. This coefficient is used to scale the original deviation threshold in real time. When the system is in a high-load long-queue state, the adjustment coefficient will automatically decrease, making the system more sensitive to small deviations. Conversely, when the queue is about to dissipate naturally, the system tolerance will relatively increase to avoid unnecessary frequent triggering. Input the AGV position deviation, environmental variable, and trigger threshold adjustment coefficient into the trigger condition judgment module to perform a composite comparison operation to judge whether the dynamic event conditions for triggering path replanning or task reassignment are met. If satisfied, immediately output a dynamic event trigger signal to provide a decision start instruction for the downstream module. To reasonably guide the direction of resource allocation, based on the current AGV position deviation, the environmental variables, and the urgency data related to task scheduling, conduct a comprehensive evaluation and construct task priority parameters. These parameters are used to quantify the response urgency of tasks and the resource allocation weights and sort them among multiple AGVs, so that AGVs in high-urgency and high-congestion risk areas can obtain path correction and task adjustment support first, thereby ensuring the stability of the operation order and scheduling efficiency of the omnidirectional AGV system in a multi-source disturbance environment.

[0027] Step 400: Input the dynamic event trigger signal and task priority parameters into the instantaneous task allocation model for solution to obtain the optimal task allocation plan and the corresponding target path set.

[0028] Specifically, based on the local environmental disturbance points marked by dynamic event trigger signals, a loading model of the lane network in the stereo garage is constructed, and based on this, the first fixed-point problem is proposed to solve the traffic distribution vector of AGVs under the current environmental constraints. This vector reflects the possible number of AGVs and the flow intensity that each lane may carry at the path execution level. The solution needs to consider factors such as channel congestion, passable boundary, and historical usage frequency of the path. A second fixed-point problem is constructed using the task priority parameter to match handling tasks of different levels to the current available AGV resources, and the task assignment vector is solved in combination with the task location, task urgency, and execution time window, so as to determine the task unit that each AGV should give priority to respond to at a specific moment. After obtaining these two core vectors, the traffic distribution vector and the task assignment vector are input into the instantaneous task assignment model. The task selection tendency of AGVs for each candidate path is evaluated through the path selection probability module. In the calculation process, factors such as historical execution success rate, path distance cost, and current passing risk are introduced, and the task selection probability data is output in the form of a probability distribution, reflecting the preference degree of the system for each path scheduling scheme. The task selection probability distribution data is input into the variational inequality solver for cost mapping calculation. This module analyzes the optimal task assignment scheme in the current system state by establishing a comprehensive cost mapping relationship including path selection cost, queue interference cost, turning loss, and passing conflict. During the solution process, under the constraint conditions of physical feasibility and scheduling consistency, the system executes the path decoding and task mapping processes to ensure that the obtained path set has connectivity in structure, uniqueness in tasks, and balance in load. Finally, the optimal task assignment scheme and the target path set that can be directly issued to the AGV system in the current garage state are obtained.

[0029] Input the traffic distribution vector and the task assignment vector into the instantaneous task assignment model, which integrates multi-factor path attribute information to achieve a comprehensive utility evaluation of candidate paths. During this process, numerical calculations of the path utility function are performed for each path that may be selected. This utility function consists of three parts: path travel time, instantaneous queuing delay, and system revenue after task completion. Among them, the path travel time reflects the actual running time required for the AGV to travel from the current position to the task point via this path. The instantaneous queuing delay is adjusted according to the current path congestion status and is used to measure the possible execution waiting time for this path. The task completion revenue represents the positive contribution of the execution of this task to the overall task completion rate, task level requirement satisfaction, or operation efficiency. After synthesizing the three items, path utility data for each candidate path is formed, serving as the basic score for subsequent probability reasoning. To adapt to the non-deterministic environment faced by task assignment in different operation scenarios, factors such as the current operation complexity, competition intensity among AGVs, task urgency, and path dynamic change frequency are considered to estimate the decision uncertainty of the AGV handling task in the stereoscopic garage during path selection, and accordingly, a probability distribution adjustment coefficient is determined. This coefficient directly affects the distribution form of the probabilities of each path being selected. After determining this adjustment coefficient, an exponential transformation calculation is performed based on the path utility data to perform an exponential mapping of the utility values of each path under the scale of the adjustment coefficient, forming utility data with exponential non-linear characteristics, thereby amplifying the advantages of superior paths and compressing the selection probabilities of sub-optimal paths, making the final path distribution more decision-biased while maintaining randomness. To ensure the normativity and mathematical validity of the output probabilities, a normalization operation is performed on the exponential utility values of all candidate paths, the exponential weights of all paths are summed up, and then the exponential value of each path is divided by the sum to obtain task selection probability distribution data that conforms to the definition of probability distribution.

[0030] In the embodiments of the present application, by constructing a dynamic environmental state matrix and calculating environmental change rate data, it is possible to accurately perceive in real time the changes in multi-dimensional environmental information such as the positions of obstacles, the states of AGVs, and the occupancy of parking spaces in a three-dimensional garage. Compared with the traditional static map method, the adaptability to complex dynamic environments is significantly improved. By establishing an instantaneous queue length calculation mechanism and a queue dissipation prediction algorithm, it is possible to accurately identify and quantify the task queues formed within a short period during the operation of a three-dimensional garage, overcoming the limitation of the prior art that only considers the average queue, and realizing the accurate modeling and processing of the instantaneous queuing effect. By designing a composite trigger condition judgment mechanism based on the AGV position deviation and environmental variables, the path replanning is only started when necessary, avoiding the continuous calculation mode of the traditional method, and significantly reducing the consumption of computing resources while ensuring the timeliness of system response. By adopting a variational inequality solution framework combined with an instantaneous task allocation model, it is possible to consider multiple decision dimensions such as AGV traffic distribution, task allocation, and path selection simultaneously, and realize the global coordinated optimization of the multi-AGV system in a three-dimensional garage. By dynamically adjusting the trigger threshold according to the queue dissipation prediction time and calculating the probability distribution based on the path utility function, the system can adaptively adjust key parameters according to the actual operation conditions, improving the robustness and adaptability of path planning.

[0031] In a specific embodiment, the process of executing step 100 may specifically include the following steps: Real-time data collection is performed on the lidar sensors, ultrasonic sensors, and vision sensors of each floor of the parking spaces in the three-dimensional garage to obtain obstacle position information, real-time AGV position information, parking space occupancy status information, and passageway passability information; A dynamic environmental state matrix is constructed based on the obstacle position information, real-time AGV position information, parking space occupancy status information, and passageway passability information; Based on the dynamic environmental state matrix, the state difference calculation within a continuous time window is performed to obtain environmental change rate data.

[0032] Specifically, a targeted sensor cluster is arranged on each floor of the three-dimensional garage, enabling each monitoring unit to not only sense the static scene state but also make a high-frequency response to dynamic event changes. Among them, the lidar sensor is responsible for constructing a three-dimensional spatial structure, actively scanning the surrounding environment on the horizontal and vertical planes through high-precision laser beams, so as to obtain spatial information such as the boundaries of obstacles in the garage, the contours of AGVs, and the wall structures. This type of data has the characteristics of high ranging accuracy and strong anti-light interference, and is suitable for capturing the contour positions of obstacles and their dynamic changes. At the same time, ultrasonic sensors are used for blind spot detection in the short-distance range. Especially in narrow channels, around AGVs, or at the edges of parking spaces, areas where lidar is blocked and unable to detect can be supplemented with stable distance data by ultrasonic sensors, which are suitable for identifying the generation of local obstacles and sudden interference sources. The vision sensor obtains video frame data through an embedded image recognition algorithm, and uses deep learning models such as YOLO or Mask R-CNN to achieve semantic information recognition such as parking space occupancy status recognition, AGV type discrimination, lane marking tracking, and lighting recognition, enabling the system to have multi-scale image recognition capabilities at the parking space level, vehicle level, and channel level. The data of the above three types of sensors are fused and preprocessed through edge computing nodes, including operations such as time synchronization, coordinate transformation, data filtering, and spatial alignment, mapping data with different precisions and dimensions into the same structured data model to form a standardized input. According to the preprocessed multi-source data, spatial reconstruction and semantic decoding are performed on the obstacle position information, AGV real-time position information, parking space occupancy status information, and channel passability information. The obstacle position information is converted into a visible closed interval of the area where the obstacle is located through point cloud clustering and boundary tracking algorithms. The AGV real-time position information is estimated by fusing methods such as IMU, odometer, and visual positioning to determine the global coordinates, orientation, and speed information of each AGV at a specific time point. The parking space occupancy status information is continuously detected by the image recognition model for the parking space contour and covered area in the video stream to determine whether there is a vehicle and generate occupancy flag data. The channel passability information is jointly judged by the above three types of data: if an obstacle, AGV staying, or insufficient lighting is detected within the channel range, the channel is marked as non-passable, otherwise it is regarded as an open channel. These information are quantified into state vectors with a consistent coding structure in the spatial coordinate system, and the set of these state vectors is the dynamic environment state matrix of the three-dimensional garage at a specific moment. Each row or column represents a monitoring unit, such as a parking space, an intersection, or a section of a channel, and the vector dimension includes multi-dimensional information such as obstacle occupancy rate, vehicle position offset, state update frequency, and passable flag. As events such as vehicle entry and exit in the garage, AGV task execution, and channel condition changes occur, this state matrix will be continuously updated over time. To achieve systematic perception of environmental dynamic changes, state difference calculations are performed within a continuous time window based on this dynamic environment state matrix.The system caches the state matrices at several consecutive moments in a time series database, constructs a state trajectory sequence, and when each new moment arrives, extracts the differences between the current matrix and the matrices at the previous moment or multiple historical moments. Through difference operations at the vector level or matrix level, the state change amplitude of each monitoring point is extracted. This difference process is not limited to numerical calculations, but also includes the switching of state classifications, such as a parking space changing from occupied to free, a passage changing from unobstructed to blocked, etc. These semantic changes are also recorded in the form of difference markers. To improve the robustness of change detection, a moving average filtering and outlier removal mechanism is introduced to prevent misjudgment caused by single-point false alarms or sudden interferences. Through the above difference operations, a set of continuous change rate data is formed, which is the core content of the environmental change rate data, reflecting the dynamic fluctuation intensity and spatial distribution pattern of the three-dimensional garage per unit time.

[0033] In a specific embodiment, the process of executing step 200 may specifically include the following steps: Based on the dynamic environmental state matrix and the environmental change rate data, the handling task flow during the operation cycle of the three-dimensional garage is statistically analyzed to obtain the task arrival rate and AGV processing capacity data; According to the task arrival rate and AGV processing capacity data, the maximum value of the instantaneous queue length is calculated to obtain the instantaneous task queue length data and the queue delay weight coefficient; Based on the instantaneous task queue length data and the queue delay weight coefficient, the task demand fluctuation intensity is calculated to generate the instantaneous task queue parameters; Based on the instantaneous task queue parameters, the queue dissipation time prediction calculation is performed to obtain the queue dissipation prediction time.

[0034] Specifically, task event trigger points are extracted from the constructed environmental state matrix, and the number of newly generated handling requests per unit time within a specific operation cycle is counted. These task events originate from car pick-up or parking instructions proposed by users, automatic adjustment instructions dispatched by the system, or task inheritance behaviors among AGVs. Moreover, each task will be reflected in the state matrix as characteristics such as changes in vehicle occupancy status, channel status switching, or sudden changes in AGV scheduling trajectories. Combining with the environmental change rate data, it is determined whether these tasks are sudden demands caused by environmental disturbances or normal task loads that occur periodically, and thus partitioned statistics are carried out according to time windows to obtain the task arrival rate, which is the basic data reflecting the task generation intensity. At the same time, by analyzing information such as the actual number of tasks completed by AGVs within the same operation cycle, the average response time, available operating time, and path reachability rate, and combining the task categories and historical performances executed by each AGV, a dynamic model of AGV processing capacity is formed to generate AGV processing capacity data corresponding to the task arrival rate. This processing capacity is not only restricted by the number of AGVs but also related to the path smoothness, traffic density, load capacity, and scheduling system efficiency. Therefore, the system conducts multi-factor calculations by integrating all dynamic environmental indicators to ensure that the processing capacity data has predictability and guiding significance in actual scheduling. The maximum value of the instantaneous queue length is calculated based on the task arrival rate and AGV processing capacity data to determine whether the task density exceeds the system processing threshold. By comparing the difference between the task generation speed and the AGV processing capacity per unit time at the current moment, the maximum instantaneous queue length that the system can bear under the current configuration and environment is estimated, and further, this queue length is associated with the historical delay time, the number of AGVs staying, and the task waiting duration to calculate the delay cost brought by each task stacking unit, and the queue delay weight coefficient is obtained. This coefficient is used to reflect the negative impact degree of task queuing on the overall system performance. The higher the weight, the more significant the impact of system delay. Based on the above data, the calculation of task demand fluctuation intensity is carried out. This process continuously monitors the change trend of the queue length within the sliding time window and combines with the queue delay weight coefficient to amplify or suppress the fluctuation amplitude to form quantifiable task demand fluctuation intensity data. This fluctuation intensity is used to describe the change in the number of tasks and also takes into account the processing imbalance caused by environmental changes and AGV congestion, thus more truly reflecting the stability and adaptability of the garage task system. On this basis, the system generates instantaneous task queue parameters, which gather multiple elements such as the current queue size, queuing penalty weight, and demand fluctuation intensity, and are the key inputs for subsequent scheduling priority determination, path adjustment, and load transfer. Using the obtained instantaneous task queue parameters, combined with the current AGV distribution, the number of available paths, and the environmental state trend, a queue dissipation time prediction model is established to simulate the natural decay trajectory of the existing task queuing state without introducing new intervention measures.Multiple influencing factors such as the average processing rate of AGVs, the degree of channel congestion, the task processing order, and the task urgency distribution are considered in the prediction calculation. By deducing the ability and speed of the system to automatically process tasks in the future time period, the time required to completely clear the expected task backlog is calculated, thereby forming the prediction time for queue dissipation.

[0035] In a specific embodiment, the process of calculating the task demand fluctuation intensity based on the instantaneous task queue length data and the queue delay weight coefficient to generate the instantaneous task queue parameters may specifically include the following steps: Perform a quantization calculation of the instantaneous queuing effect on the instantaneous task queue length data and the queue delay weight coefficient to obtain the queuing effect quantization data; Based on the queuing effect quantization data, perform an analysis of the historical task arrival rate sequence to obtain the average value and standard deviation data of the task arrival rate; Based on the average value and standard deviation data of the task arrival rate, perform a variance calculation of the task demand fluctuation intensity to obtain the task demand fluctuation intensity data and the comparison result with the fluctuation threshold; According to the task demand fluctuation intensity data and the comparison result with the fluctuation threshold, perform an instantaneous queuing effect determination to generate the instantaneous task queue parameters.

[0036] Specifically, perform instantaneous queuing effect quantification calculation on the instantaneous task queue length data and the queue delay weight coefficient. Obtain the length information of the AGV task queue within each time sampling period. The queue length reflects the backlog scale of the unfinished tasks in the system, while the delay weight coefficient is obtained by statistically analyzing the impact of the task completion time delay on the overall performance in the early stage. The higher the weight, the more significant the impact of queuing on the overall system efficiency. Multiply the queue length by the delay weight to obtain the queuing effect quantification data. Based on the queuing effect quantification data, perform historical task arrival rate sequence analysis to obtain a stable reference value as the baseline for volatility judgment. This analysis retrospectively examines the task trigger records within a certain time range, counts the number of task arrivals in each time slice to form a time series data set, and calculates the mean value and extracts the standard deviation of this set to obtain the central tendency and dispersion degree of the task arrival rate respectively. The mean value represents the average load level of the system under stable operating conditions, while the standard deviation reflects the variation range of the task generation frequency and is an important reference for measuring whether the system is in a state of periodic fluctuation, abnormal surge, or load mutation. Perform variance calculation of the task demand fluctuation intensity based on the mean value and standard deviation data of the task arrival rate to quantify the difference degree between the current task system operating state and the long-term average state. The calculation method of this fluctuation intensity is to square the deviation between the actual task arrival rate in the current time period and the historical mean value, and normalize it in combination with the standard deviation ratio to obtain a numerical expression that can truly reflect the severity of the current load fluctuation of the system. The introduction of this fluctuation intensity data can reveal whether the system is in a short-term shock or a long-term anomaly, and accordingly set a reasonable response strategy. After obtaining the task demand fluctuation intensity data, compare this value with a preset fluctuation threshold. The fluctuation threshold is jointly determined by the maximum acceptable load change range, scheduling fault tolerance ability, and response delay tolerance in historical operation and is the demarcation benchmark for judging whether the system needs to enter rescheduling or path reconstruction. Complete the final determination of the current instantaneous queuing effect according to the comparison result between the task demand fluctuation intensity data and the threshold. If the fluctuation intensity exceeds the threshold, it indicates that the matching state between the current task generation and the system processing ability has been seriously imbalanced. At this time, the system marks the current state as an abnormal load state or a high-risk queuing state and generates new instantaneous task queue parameters accordingly. This parameter includes the instantaneous queue length, delay impact coefficient, fluctuation intensity value, determination label, and subsequent trigger priority.

[0037] In a specific embodiment, the process of performing step 300 may specifically include the following steps: Calculate the Euclidean distance between the actual position and the planned position of the AGV to obtain the AGV position deviation, and perform weighted summation of environmental variables based on the instantaneous task queue parameters to obtain environmental variables; The adaptive threshold is dynamically adjusted according to the ratio of the queue dissipation prediction time to the operation cycle time to obtain a trigger threshold adjustment coefficient; Based on the AGV position deviation, environmental variables, and trigger threshold adjustment coefficient, a composite trigger condition comparison and judgment are performed to obtain a dynamic event trigger signal; The AGV position deviation, environmental variables, and task urgency data are evaluated for priority to obtain task priority parameters.

[0038] Specifically, a dynamic event-triggering mechanism is constructed with the monitoring of AGV execution deviation as the main line, combined with environmental changes, system load, and task urgency as criteria, and it is ensured that this mechanism has the comprehensive perception ability of operation deviation, traffic complexity, and scheduling pressure. Calculate the Euclidean distance between the actual position and the planned position of the AGV to obtain the spatial deviation value of the current position of the AGV relative to the ideal path. The larger this position deviation value is, the higher the degree of interference, blockage, or path instability that the AGV experiences during task execution, which is an important precursor to the decline of system scheduling efficiency or the increase of path failure risk. At the same time, the environmental factors are weighted and aggregated in combination with the current instantaneous task queue parameters to construct an environmental variable that can reflect the degree of external pressure and environmental interference of the system. In this process, factors such as the instantaneous task queue length, task arrival density, channel traffic impedance, and queuing delay weight are used as basic inputs, and a weighted summation model is used to integrate them into a single environmental complexity index. This environmental variable can comprehensively describe the traffic congestion degree, resource competition intensity, and task conflict density within the path or area where the current AGV is located. Since the environmental variable is continuously affected by dynamic task fluctuations and AGV position distributions, this calculation process needs to be updated periodically to ensure that each trigger judgment is based on the latest data for decision-making. To enable the trigger mechanism to adapt to different operation rhythms and load intensities, a dynamic threshold adjustment mechanism based on the ratio relationship between the predicted queue dissipation time and the standard operation cycle time is introduced. If the system predicts that the current task backlog state will naturally resolve in the short term, it means that there is no need to immediately perform path reconstruction or task reallocation. Therefore, the system automatically relaxes the trigger threshold to avoid unnecessary scheduling operations caused by small deviations. Conversely, if the prediction shows that the task backlog persists or even deteriorates further, the system will lower the trigger threshold to increase the sensitivity to operation deviations and environmental anomalies, so as to intervene early for scheduling adjustments. This adaptive threshold adjustment process will dynamically generate a trigger threshold adjustment coefficient according to the proportional coefficient between the predicted queue dissipation time and the operation beat cycle output by the prediction model, and use it for the standard correction of subsequent judgment conditions. Input the AGV position deviation, environmental variable, and trigger threshold adjustment coefficient into the composite condition comparison module to perform multi-dimensional fusion judgment. The specific logic is to compare the weighted sum of the position deviation value and the environmental variable with the adjusted dynamic trigger threshold. If the weighted result exceeds the threshold, it is determined that the current system operation has deviated from the normal scheduling trajectory, and a dynamic event trigger signal needs to be generated immediately. Input this signal into the task allocation model or path replanning module to start the system response chain. To implement the resource priority scheduling mechanism, a joint analysis is performed on the above-mentioned calculated AGV position deviation, environmental variable, and the urgency data of each task to construct a task priority evaluation model. In this model, the task urgency is determined by multiple factors, including the time interval between the task request time and the current moment, the importance of the task target vehicle, the user reservation time limit, or the garage access flow control strategy, etc.The system uniformly quantifies three types of input data and assigns different evaluation weights, performs multi-factor linear or non-linear combinations, and forms a numerical task priority parameter.

[0039] In a specific embodiment, the process of executing step 400 may specifically include the following steps: Construct a first fixed-point problem for garage road network loading based on the dynamic event trigger signal and solve the AGV traffic distribution to obtain a traffic distribution vector. Construct a second fixed-point problem for garage road network loading based on the task priority parameter and solve the task allocation vector to obtain a task allocation vector; Calculate the path selection probability for the traffic distribution vector and the task allocation vector through the instantaneous task allocation model to obtain task selection probability distribution data; Input the task selection probability distribution data into the variational inequality solver for cost mapping calculation, and at the same time perform path decoding and task allocation mapping calculation under the constraint conditions of the feasible solution set to obtain the optimal task allocation plan and the corresponding target path set.

[0040] Specifically, based on the dynamic event trigger signal, identify the key areas or key nodes in the current three-dimensional garage network that require path replanning or task adjustment, and build a structured model of the garage road network based on this. This road network model converts the spatial distribution of current AGVs, the set of passable paths, the channel status, the parking space nodes, and the path connectivity relationship into a network graph structure, where each edge represents an actual channel, each node represents a task target or an AGV state transition point, and the weight of the edge represents composite information such as traffic cost, delay risk, or traffic pressure. When the dynamic event trigger signal is determined by the system to be valid, the system constructs a first fixed-point problem based on the current trigger location and influence range, thereby analyzing the AGV traffic intensity carried by each channel in the current garage network and forming a dynamic mapping of the local area traffic load. To this end, the system loads the location information, target point information, and traffic status information of the current AGV in the graph model, takes the distribution of AGVs in all possible paths as variables, and establishes a problem of solving the AGV traffic distribution in the current state by setting constraints on path load, reachability, trafficability, and blockage status. Then, through iterative approximation or network equilibrium algorithms, the current traffic distribution vector of each channel is obtained. This vector is used to describe the resource occupancy level of AGVs on each path and is an important basis for subsequent task scheduling weights and path conflict judgments. Based on the obtained traffic distribution vector, a second fixed-point problem is constructed according to the task priority parameters. This problem takes the urgency of task requirements, the resource competition status, and the availability of AGVs as inputs and aims to maximize the task assignment matching degree to solve the optimal task assignment vector. In the modeling process, each task is regarded as a resource unit to be assigned, each AGV is the task executor, and the task priority parameter is used as the weight coefficient for each pair of AGV-task combinations to express the quality of their matching in the current environment. The system constructs a feasible assignment plan through the edges between AGV nodes and task nodes in the graph structure, and performs optimization calculations under the conditions of meeting traffic constraints and task uniqueness to obtain the task assignment vector, the content of which characterizes which AGV is most likely to undertake and execute each task in what path manner. The traffic distribution vector and the task assignment vector are jointly input into the instantaneous task assignment model for reasoning and calculation of path selection probabilities. This model is based on the utility value of each candidate path, constructs a path utility function by combining the path travel time, the current traffic, the delay risk, and the system benefits brought by task completion, and adopts an exponential mapping strategy to calculate the selection probability of each path under the task-AGV assignment combination considering the task assignment relationship. This probability reflects the preference degree of the system for scenarios with multiple paths coexisting and intense resource competition in the current environmental state and is a key control quantity in the non-deterministic scheduling system. The system normalizes the path selection probabilities to a probability distribution data with a sum of one. The task selection probability distribution data is input into the variational inequality solver for cost mapping calculation.The solver combines the energy consumption, time consumption, blocking risk, queuing effect and resource competition on the task execution path into a unified cost function by constructing a system cost mapping relationship, thereby transforming the task allocation problem into an optimal solution problem with inequality constraints. The variational inequality solution allows the system to find the optimal scheduling state under nonlinear and multi-constraint conditions, avoiding the defects of traditional linear optimization methods that fail or have low computational efficiency in complex environments. In this process, the system simultaneously loads the feasible solution set, that is, the full set of all legal path combinations and task-AGV matching combinations, and performs path decoding operations in the set, that is, numbering and binding specific paths according to the selection probability, and then performing task mapping operations based on the task allocation vector and path encoding data, thereby completing the mapping encapsulation from task requirements to path resources in mathematical structure. The system outputs the optimal task allocation plan and the corresponding target path set on the premise of satisfying all path access constraints, task allocation constraints and load balancing constraints. The optimal task allocation plan describes which tasks each AGV should undertake, and the target path set contains the physical path sequence and control instructions corresponding to each task for the scheduling system to send to the AGV controller for execution.

[0041] In a specific embodiment, the execution step calculates the path selection probability of the traffic distribution vector and the task allocation vector through the instantaneous task allocation model to obtain the task selection probability distribution data, which may specifically include the following steps: The traffic distribution vector and the task allocation vector are input into the instantaneous task allocation model, and the path utility function value of each candidate path is calculated by the path travel time, instantaneous queue delay and task completion benefit to obtain the path utility data; Determine the probability distribution adjustment coefficient according to the decision uncertainty degree of the AGV handling task in the stereo garage; Based on the path utility data and the probability distribution adjustment coefficient, an exponential transformation calculation is performed to obtain the indexed utility value of each candidate path; The indexed utility values of each candidate path are normalized by probability calculation to obtain the task selection probability distribution data.

[0042] Specifically, taking the AGV traffic distribution vector and the task assignment vector as the input basis, they respectively represent the AGV usage density per unit time in each channel of the current multi-story garage and the execution resource allocation relationship bound to each task. The instantaneous task assignment model comprehensively evaluates each candidate path in the optional path set in turn after receiving these two vectors. The main content of the evaluation focuses on three dimensions: First, the path passing time, which is comprehensively predicted through the historical path execution duration, the current traffic saturation state, and the local path topology. The shorter it is, the less time the AGV needs to reach the target task point from the current position; Second, the instantaneous queuing delay, which is estimated in real time based on the number of AGVs on the current path, the available width of the channel, the number of conflict nodes, and the queuing depth model. It is a quantitative compensation for non-expected events such as waiting, deceleration, and replanning that the AGV may encounter during execution; Third, the task completion benefit, which comes from the system target gain after the task is completed in the task priority parameter, such as the contribution to the garage access efficiency, the shortening of the user waiting time, or the improvement of the system balance rate. The higher it is, the greater the impact of the task corresponding to the path on the global state of the system. The three indicators are weighted and combined to form the path utility function value, that is, the path utility data, which establishes a numerical scoring system for each path in the current environment and scheduling state, thereby reflecting the comprehensive attractiveness of the path in the scheduling selection. The higher the utility value, the better the comprehensive performance of the path in terms of time efficiency, smooth passage, and system benefits. Considering the uncertainties such as multiple path options, multiple task competitions, and multiple AGV concurrency in the actual scheduling process, a probability distribution mechanism is introduced to express the path selection tendency in a non-deterministic form. For this purpose, the decision-making uncertainty of the AGV handling task in the multi-story garage is evaluated, and a probability distribution adjustment coefficient is determined accordingly. This coefficient, as the steepness adjustment factor of the path selection probability curve, can control the trade-off between the scheduling system in "tending to the optimal path" and "retaining the randomness of multiple paths". If the uncertainty is high, for example, there are a large number of temporary tasks in the system, the path state changes frequently, and there are uncontrollable factors in the AGV state, the adjustment coefficient should be set to a lower value to maintain the dispersion and redundancy of path selection; conversely, if the system state is stable, the path is smooth, and the scheduling goal is clear, the adjustment coefficient should be set to a higher value accordingly, so as to increase the selection tendency for the optimal path. Perform an exponential transformation calculation on the aforementioned path utility data with the adjustment coefficient, take the utility value of each path as the input of the exponential function, the independent variable is the path utility value, and the exponential function is scaled based on the adjustment coefficient, so as to amplify the advantages of high-utility paths and suppress the probability weight of low-utility paths. The transformation result forms a set of exponential utility values, which reflects the scheduling attractiveness of each path under the control of the adjustment factor. Normalize the exponential utility values of all candidate paths.Sum all the path index values, and then divide the exponentiated value of each path by this sum to ensure that the sum of the selection probabilities of all paths is one, obtaining the task selection probability distribution data that satisfies the probability definition.

[0043] In this embodiment, the task selection probability distribution data is input into a variational inequality solver for cost mapping calculation, and at the same time, path decoding and task assignment mapping calculations are performed under the constraint conditions of the feasible solution set, obtaining the optimal task assignment scheme and the corresponding target path set, including: constructing a variational inequality mathematical expression based on the task selection probability distribution data, establishing a variational inequality problem by finding the optimal solution vector that satisfies the inequality constraint conditions, and obtaining the variational inequality constraint conditions; performing a mapping function calculation on the path cost including the instantaneous queuing effect according to the variational inequality constraint conditions to obtain the cost mapping function value; performing an intersection operation on the cost mapping function value and the three-dimensional garage path constraint conditions to obtain the boundary range of the feasible solution set; performing iterative calculations based on the boundary range of the feasible solution set using an adaptive step size continuous weighted average algorithm, updating the next round of iterative solution through the weighted combination of the current solution vector and the historical solution vector to obtain a convergent solution vector; performing path decoding and task assignment mapping calculations on the convergent solution vector to obtain the optimal task assignment scheme and the corresponding target path set.

[0044] The path planning method of the omnidirectional AGV in the embodiment of the present application has been described above. Next, the path planning device of the omnidirectional AGV in the embodiment of the present application will be described. Please refer to Figure 2 , an embodiment of the path planning device of the omnidirectional AGV in the embodiment of the present application includes: An acquisition module 11, configured to acquire the dynamic environment state matrix and the environmental change rate data in the three-dimensional garage; A calculation module 12, configured to perform an instantaneous queue length calculation on the task arrival rate and the AGV processing capacity data during the operation cycle of the three-dimensional garage according to the dynamic environment state matrix and the environmental change rate data, obtaining the instantaneous task queue parameters and the queue dissipation prediction time; A judgment module 13, configured to perform a composite trigger condition judgment on the AGV position deviation and the environmental variables based on the instantaneous task queue parameters and the queue dissipation prediction time, obtaining a dynamic event trigger signal and task priority parameters; A solution module 14, configured to input the dynamic event trigger signal and the task priority parameters into an instantaneous task assignment model for solution, obtaining the optimal task assignment scheme and the corresponding target path set.

[0045] Through the collaborative cooperation of the above-mentioned various components, by constructing a dynamic environmental state matrix and calculating environmental change rate data, it is possible to accurately perceive in real time the changes in multi-dimensional environmental information such as the positions of obstacles in the stereoscopic garage, the states of AGVs, and the occupancy of parking spaces. Compared with the traditional static map method, the adaptability to complex dynamic environments is significantly improved. Establishing an instantaneous queue length calculation mechanism and a queue dissipation prediction algorithm can accurately identify and quantify the task queues formed within a short time during the operation of the stereoscopic garage, overcoming the limitations of the existing technology that only considers the average queue, and realizing the precise modeling and processing of the instantaneous queuing effect. Designing a composite trigger condition judgment mechanism based on the position deviation of the AGV and environmental variables to start path replanning only when necessary, avoiding the continuous calculation mode of traditional methods, and significantly reducing the consumption of computing resources while ensuring the timeliness of system response. Adopting a variational inequality solution framework combined with an instantaneous task allocation model can simultaneously consider multiple decision dimensions such as AGV traffic distribution, task allocation, and path selection, and achieve the global coordinated optimization of the multi-AGV system in the stereoscopic garage. By dynamically adjusting the trigger threshold according to the queue dissipation prediction time and calculating the probability distribution based on the path utility function, the system can adaptively adjust key parameters according to the actual operation conditions, improving the robustness and adaptability of path planning.

[0046] Please refer to Figure 3 , Figure 3 FIG. is a schematic block diagram of the structure of the electronic device 300 provided by an embodiment of the present application. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected through a device bus 303. Among them, the memory 302 may include a non-volatile storage medium and an internal memory.

[0047] The non-volatile storage medium can store a computer program. The computer program includes program instructions. When the program instructions are executed by the processor 301, the processor 301 can be enabled to execute any of the above-mentioned path planning methods for omnidirectional AGVs.

[0048] The processor 301 is used to provide computing and control capabilities to support the operation of the entire electronic device 300.

[0049] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can be enabled to execute any of the above-mentioned path planning methods for omnidirectional AGVs.

[0050] Those skilled in the art can understand that Figure 3 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the electronic device 300 involved in the solution of the present application. The specific electronic device 300 may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0051] It should be understood that the processor 301 can be a Central Processing Unit (CPU), and the processor 301 can also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0052] It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described electronic device 300 can refer to the corresponding process of the path planning method of the aforementioned omnidirectional AGV, which will not be elaborated here.

[0053] The embodiment of the present application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program. When the computer program is executed by one or more processors, the one or more processors are caused to implement the path planning method of the omnidirectional AGV provided by the embodiment of the present application.

[0054] Among them, the computer-readable storage medium can be the internal storage unit of the electronic device 300 in the foregoing embodiment, such as the hard disk or memory of the electronic device 300. The computer-readable storage medium can also be an external storage device of the electronic device 300, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped with the electronic device 300.

[0055] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, which will not be elaborated here.

[0056] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing an electronic device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0057] As described above, the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit them; although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of this application.

Claims

1. A path planning method for an omnidirectional AGV, characterized in that, Including: Obtain the dynamic environment state matrix and environmental change rate data in the stereo garage; Perform instantaneous queue length calculation on the task arrival rate and AGV processing capacity data within the operation cycle of the stereo garage according to the dynamic environment state matrix and the environmental change rate data, to obtain instantaneous task queue parameters and queue dissipation prediction time; Based on the instantaneous task queue parameters and the queue dissipation prediction time, judge the composite trigger conditions for AGV position deviation and environmental variables, to obtain dynamic event trigger signals and task priority parameters; Input the dynamic event trigger signals and the task priority parameters into the instantaneous task allocation model for solution, to obtain the optimal task allocation scheme and the corresponding target path set.

2. The path planning method of the omnidirectional AGV according to claim 1, wherein The obtaining of the dynamic environment state matrix and environmental change rate data in the stereo garage includes: Perform real-time data collection on the lidar sensors, ultrasonic sensors and vision sensors of each floor parking space in the stereo garage, to obtain obstacle position information, AGV real-time position information, parking space occupancy status information and passage passability information; Construct a dynamic environment state matrix according to the obstacle position information, the AGV real-time position information, the parking space occupancy status information and the passage passability information; Based on the dynamic environment state matrix, perform state difference calculation within a continuous time window, to obtain environmental change rate data.

3. The path planning method of the omnidirectional AGV according to claim 1, characterized in that The performing of instantaneous queue length calculation on the task arrival rate and AGV processing capacity data within the operation cycle of the stereo garage according to the dynamic environment state matrix and the environmental change rate data, to obtain instantaneous task queue parameters and queue dissipation prediction time, includes: Statistically analyze the handling task flow within the operation cycle of the stereo garage based on the dynamic environment state matrix and the environmental change rate data, to obtain the task arrival rate and AGV processing capacity data; Perform maximum calculation of the instantaneous queue length according to the task arrival rate and the AGV processing capacity data, to obtain instantaneous task queue length data and queue delay weight coefficient; Perform task demand fluctuation intensity calculation based on the instantaneous task queue length data and the queue delay weight coefficient, to generate instantaneous task queue parameters; Perform queue dissipation time prediction calculation based on the instantaneous task queue parameters, to obtain queue dissipation prediction time.

4. The path planning method of the omnidirectional AGV according to claim 3, characterized in that, The performing of task demand fluctuation intensity calculation based on the instantaneous task queue length data and the queue delay weight coefficient, to generate instantaneous task queue parameters, includes: Perform instantaneous queuing effect quantization calculation on the instantaneous task queue length data and the queue delay weight coefficient, to obtain queuing effect quantization data; Perform historical task arrival rate sequence analysis based on the queuing effect quantization data, to obtain task arrival rate average value and standard deviation data; Perform task demand fluctuation intensity variance calculation based on the task arrival rate average value and the standard deviation data, to obtain task demand fluctuation intensity data and fluctuation threshold comparison result; Perform instantaneous queuing effect determination according to the task demand fluctuation intensity data and the fluctuation threshold comparison result, to generate instantaneous task queue parameters.

5. The path planning method of the omnidirectional AGV according to claim 1, characterized in that, Judging the compound trigger conditions of the AGV position deviation and the environmental variables based on the instantaneous task queue parameters and the queue dissipation prediction time to obtain a dynamic event trigger signal and task priority parameters, including: Calculating the Euclidean distance between the actual position and the planned position of the AGV to obtain the AGV position deviation, and performing weighted summation of the environmental variables based on the instantaneous task queue parameters to obtain the environmental variables; Performing adaptive threshold dynamic adjustment according to the ratio of the queue dissipation prediction time to the operation cycle time to obtain a trigger threshold adjustment coefficient; Performing a compound trigger condition comparison and judgment based on the AGV position deviation, the environmental variables, and the trigger threshold adjustment coefficient to obtain a dynamic event trigger signal; Evaluating the priorities of the AGV position deviation, the environmental variables, and the task urgency data to obtain task priority parameters.

6. The path planning method of the omnidirectional AGV according to claim 1, characterized in that, Inputting the dynamic event trigger signal and the task priority parameters into an instantaneous task allocation model for solution to obtain an optimal task allocation scheme and a corresponding target path set, including: Constructing a first fixed-point problem for loading the garage road network based on the dynamic event trigger signal and solving the AGV traffic distribution to obtain a traffic distribution vector, constructing a second fixed-point problem for loading the garage road network based on the task priority parameters and solving the task allocation vector to obtain a task allocation vector; Calculating the path selection probability of the traffic distribution vector and the task allocation vector through an instantaneous task allocation model to obtain task selection probability distribution data; Inputting the task selection probability distribution data into a variational inequality solver for cost mapping calculation, and simultaneously performing path decoding and task allocation mapping calculation under the constraint conditions of the feasible solution set to obtain an optimal task allocation scheme and a corresponding target path set.

7. The path planning method of the omnidirectional AGV according to claim 6, characterized in that Calculating the path selection probability of the traffic distribution vector and the task allocation vector through an instantaneous task allocation model to obtain task selection probability distribution data, including: Inputting the traffic distribution vector and the task allocation vector into an instantaneous task allocation model, and calculating the path utility function values of each candidate path through the path passing time, the instantaneous queuing delay, and the task completion benefit to obtain path utility data; Determining a probability distribution adjustment coefficient according to the decision uncertainty degree of the AGV handling task in the three-dimensional garage; Performing exponential transformation calculation based on the path utility data and the probability distribution adjustment coefficient to obtain the exponential utility values of each candidate path; Performing a normalized probability operation on the exponential utility values of each candidate path to obtain task selection probability distribution data.

8. A path planning device for an omnidirectional AGV, characterized in that, For executing the path planning method of the omnidirectional AGV according to any one of claims 1-7, the path planning device of the omnidirectional AGV includes: An acquisition module, configured to acquire a dynamic environment state matrix and environmental change rate data in the three-dimensional garage; A calculation module, configured to calculate the instantaneous queue length of the task arrival rate and the AGV processing capacity data during the operation cycle of the three-dimensional garage according to the dynamic environment state matrix and the environmental change rate data to obtain instantaneous task queue parameters and queue dissipation prediction time; A judgment module, configured to perform a composite trigger condition judgment on the AGV position deviation and environmental variables based on the instantaneous task queue parameters and the queue dissipation prediction time, and obtain a dynamic event trigger signal and task priority parameters; A solution module, configured to input the dynamic event trigger signal and the task priority parameters into an instantaneous task allocation model for solution, and obtain an optimal task allocation scheme and a corresponding set of target paths.

9. An electronic device, characterized in that, The electronic device includes: a memory and at least one processor, and instructions are stored in the memory; The at least one processor calls the instructions in the memory, so that the electronic device executes the path planning method for an omnidirectional AGV according to any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that, When the instructions are executed by the processor, the path planning method for an omnidirectional AGV according to any one of claims 1-7 is implemented.

Images