An AGV warehouse entry trolley and a warehouse entry path planning method thereof

Abstract

The application discloses an AGV warehousing trolley and a warehousing path planning method thereof, and belongs to the technical field of warehousing logistics, comprising a scheduling planning module, a perception reconstruction module, a control navigation module, a communication coordination module, a resource allocation module, a positioning adjustment module, a leveling execution module, a recycling energy storage module and an energy management module; the application improves the efficiency and robustness of the path, realizes conflict detection and priority scheduling among the AGV warehousing trolleys, reduces congestion and collision, and ensures stable and reliable warehousing actions; meanwhile, the application ensures the feasibility and execution stability of the fine-tuning path, realizes dynamic, reasonable and efficient allocation and recycling of resources, improves the overall scheduling efficiency, and further guarantees the task continuity.

Description

Technical Field

[0001] This invention relates to the field of warehousing and logistics technology, and in particular to an AGV (Automated Guided Vehicle) for inbound transport and its inbound route planning method. Background Technology

[0002] With the development of intelligent logistics and intelligent warehousing technologies, automated guided vehicles (AGVs) have become one of the important pieces of equipment for realizing the automation and intelligence of material handling in modern warehousing and logistics systems. Especially in intelligent warehouses, AGVs undertake key tasks such as cargo handling, inbound and outbound operations. Traditional AGV path planning methods are mostly based on static maps, and the paths do not have dynamic perception capabilities. This can easily lead to problems such as path congestion, resource conflicts, and poor docking accuracy in complex environments. At the same time, the allocation and management of resources is relatively crude, lacking comprehensive consideration of task urgency, vehicle status, and historical resource usage. In addition, during the rack docking process, due to environmental changes and image errors, AGVs often have difficulty accurately completing the inbound action, affecting operational efficiency and accuracy.

[0003] Existing AGV inbound vehicles and their inbound path planning methods suffer from low path planning efficiency and lack coordination of AGV inbound vehicles. Furthermore, existing AGV inbound vehicles and their inbound path planning methods have insufficient path fine-tuning accuracy and a single resource management approach. Therefore, we propose an AGV inbound vehicle and its inbound path planning method. Summary of the Invention

[0004] The purpose of this invention is to address the deficiencies in the existing technology by proposing an AGV (Automated Guided Vehicle) for warehouse entry and its entry path planning method.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] An AGV (Automated Guided Vehicle) for warehouse entry includes a scheduling and planning module, a perception and reconstruction module, a control and navigation module, a communication and coordination module, a resource allocation module, a positioning and adjustment module, a leveling and execution module, a recovery and energy storage module, and an energy management module.

[0007] The scheduling and planning module is used to construct a dynamic weight matrix and plan the basic path for each AGV to move into the warehouse.

[0008] The perception reconstruction module is used to acquire the pose information of each AGV entering the warehouse and the surrounding environment information in real time, so as to dynamically generate the driving path.

[0009] The control and navigation module is used to calculate the wheel speed and steering commands of each AGV inbound vehicle and control each AGV inbound vehicle to avoid obstacles in real time.

[0010] The communication coordination module is used to build the communication foundation for the coupled scheduling of task-energy-traffic flow and generate resolution strategies;

[0011] The resource allocation module is used to dynamically allocate resource usage permissions and time quotas based on the reservation requests sent by each AGV inbound vehicle.

[0012] The positioning adjustment module is used to acquire the visual features of the shelf and calculate the pose deviation between each AGV inbound trolley and the target storage location.

[0013] The leveling execution module is used to control each AGV inbound trolley to perform dynamic leveling and to perform pallet storage and retrieval operations.

[0014] The energy recovery and storage module is used to convert waste heat into electrical energy and store the generated electrical energy;

[0015] The energy management module is used to preheat or assist in cooling the battery packs of each AGV entering the warehouse.

[0016] As a further aspect of the present invention, the specific steps of the perception reconstruction module in dynamically generating the driving path are as follows:

[0017] S1.1: Obtain a static map of the warehouse, using aisle intersections, shelf gaps, and charging station locations as nodes, and connecting aisles between nodes as edges. Then, obtain the starting point and target storage area from the task instructions. Simultaneously, calculate the actual cost from the starting point to the corresponding node and the Manhattan distance from each node to the target storage location to generate a basic path. The specific formula for calculating the actual cost is as follows:

[0018]

[0019] In the formula, Indicates the actual cost, Indicates the first in the path The weight coefficient of the segment edge, Indicates the first in the basic path The actual physical length of the segment edge;

[0020] The specific formula for calculating the Manhattan distance is as follows:

[0021]

[0022] In the formula, Indicates distance from Manhattan. Represents a node Coordinates in the warehouse plane coordinate system Indicates the target storage area The coordinates of the reference point;

[0023] S1.2: Smooth the generated basic path, and then collect the current position of each AGV inbound vehicle, the congestion of the surrounding channels, the coordinates of dynamic obstacles, the battery level of each AGV inbound vehicle, and the task urgency update in real time, while obtaining the parameters of the dynamic weight matrix.

[0024] S1.3: Divide the basic path into multiple micro-path segments according to a preset interval, calculate the comprehensive score of each micro-path segment, mark each micro-path segment with a score lower than a preset threshold as a segment to be optimized, and search for each connected path within a preset range around each segment to be optimized to generate multiple candidate micro-path segments.

[0025] S1.4: Calculate the resilience value of each candidate micropath segment, rank them from highest to lowest, select the top-ranked candidate micropath segment as the alternative path, embed the alternative path into the base path, and simultaneously use a smoothing algorithm to correct the connection points between the alternative path and the base path to generate a complete path. The specific formula for calculating the resilience value is as follows:

[0026]

[0027] In the formula, Indicates the toughness value. Indicates the width adaptation factor. Represents the historical stability coefficient. This represents the time compatibility coefficient.

[0028] As a further aspect of the present invention, the specific steps for the communication coordination module to generate the resolution strategy are as follows:

[0029] S2.1: Extract each node in the complete path in real time, predict the time of each path node, generate the corresponding expected arrival time window, obtain the reservation information of each AGV inbound vehicle, and synchronize it to the external central dispatch system and adjacent AGV inbound vehicles in the path overlap area via wireless network.

[0030] S2.2: The receiver writes the information into the local path reservation cache table, compares the path node overlap, and if multiple AGV inbound vehicles appear at the same time window on the same path node, then conflict prediction is performed, and the conflict level of the current AGV inbound vehicle is judged according to the preset division rules as high level, medium level and low level.

[0031] S2.3: Calculate the corresponding resource priority score based on the task urgency, energy criticality, path substitutability, path resilience, and historical waiting factor of each conflicting AGV inbound vehicle, and sort them from high to low. At the same time, retain the original path of the AGV inbound vehicle with a score higher than the preset threshold according to the ranking order, and adjust the path of other AGV inbound vehicles.

[0032] As a further aspect of the present invention, the specific steps for the positioning adjustment module to calculate the pose deviation between each AGV inbound trolley and the target cargo location are as follows:

[0033] S3.1: Collect image data of the front structure of the shelf, obtain coordinate information based on the image data of the front structure of each shelf, and then convert the image coordinates into the spatial pose of each AGV inbound trolley in the coordinate system through calibration parameters, so as to output the corresponding three-dimensional coarse positioning information.

[0034] S3.2: Obtain the docking posture of each AGV relative to the cargo location in history, and compare it with the self-pose of each AGV inbound vehicle and the pose of the cargo location in the 3D coarse positioning information. Calculate the pose deviation between each AGV inbound vehicle and the target cargo location. The specific formula for calculating the pose deviation is as follows:

[0035] The specific formula for calculating position deviation is as follows:

[0036]

[0037] In the formula, This indicates the positional deviation. Indicates the actual position of the AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, Indicates the actual position of each AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate;

[0038] The specific formula for calculating attitude angle deviation is as follows:

[0039]

[0040] In the formula, Indicates the attitude angle deviation. This indicates the actual attitude angle of each AGV entering the warehouse. This represents the expected attitude angle of each AGV entering the warehouse;

[0041] S3.3: Compare the calculated pose deviation with the target pose of the storage location, calculate the compensation adjustment required for each AGV inbound trolley, and make minor corrections to the pose of each AGV inbound trolley based on the calculated compensation adjustment. The specific formula for calculating the compensation adjustment is as follows:

[0042]

[0043] In the formula, This represents the pose compensation adjustment amount, which includes three components, corresponding to X-axis position compensation, Y-axis position compensation, and attitude angle compensation, respectively. This indicates the X and Y axis position compensation ratio coefficients. This represents the attitude angle compensation ratio coefficient. The X and Y coordinates represent the orientation of the target cargo location. The current actual X and Y coordinates of the AGV inbound vehicle. Target attitude angle of the cargo location The current actual attitude angle of each AGV entering the warehouse.

[0044] A method for planning the inbound path of an AGV (Automated Guided Vehicle) inbound vehicle, the specific steps of which are as follows:

[0045] Ⅰ: Collect and parse the inbound task information, and generate the basic path for each AGV inbound vehicle based on the static warehouse map;

[0046] II: Real-time collection of the location and surrounding environment information of each AGV entering the warehouse, evaluation and adjustment of the generated basic path;

[0047] Ⅲ: Collect the path reservation information sent by each AGV inbound vehicle, predict conflicts based on the path reservation information, and coordinate various conflicts.

[0048] IV: Collect real-time resource occupancy data and resource usage requests and reservations from each AGV (Automated Guided Vehicle) for inbound trolleys, and dynamically allocate quotas;

[0049] V: Calculate the pose deviation between each AGV inbound trolley and the target storage location to generate the corresponding fine-tuning path.

[0050] As a further aspect of the present invention, the specific steps for dynamic quota allocation in step IV are as follows:

[0051] S4.1: Collect and assign corresponding identifiers to various management resources, detect in real time whether there are AGV inbound vehicles occupying resources in the management resource area, and retrieve the occupancy records of various resources within 1 hour from the local resource database, and calculate the average occupancy time and peak usage period.

[0052] S4.2: Collect resource reservation requests from each AGV inbound vehicle, arrange each resource reservation request in ascending order, mark each resource reservation request as pending, evaluate the priority of each resource reservation request, and arrange them in descending order.

[0053] S4.3: Based on the resource occupancy status, prioritize each resource reservation request, allocate usage quotas for each type of resource, generate corresponding quota instructions, and then send the quota instructions to the corresponding AGV inbound vehicle. If the AGV inbound vehicle with the allocated quota malfunctions or the task is canceled, the originally allocated quota is reclaimed and reassigned to the AGV inbound vehicle with the highest priority of the resource reservation request.

[0054] As a further aspect of the present invention, the specific steps for generating the corresponding fine-tuning path in step V are as follows:

[0055] S5.1: Collect the reference pose data of the target location and the image feature points of the shelf, extract the pixel coordinates of the shelf image feature points, correct the pixel coordinates of the shelf image feature points through the distortion coefficient, and then convert the corrected pixel coordinates into coordinates in the three-dimensional coordinate system through the perspective projection formula.

[0056] S5.2: Convert the coordinates in the three-dimensional coordinate system to the coordinates in the AGV loading vehicle coordinate system to obtain the current planar position of each AGV loading vehicle and calculate the pose deviation between each AGV loading vehicle and the target storage location. At the same time, set the constraint parameters for fine-tuning the path based on the physical performance of each AGV loading vehicle and the environment of the target storage location.

[0057] S5.3: Based on the calculated pose deviation and the constraint parameters of the set fine-tuning path, a corresponding fine-tuning path is generated, and each fine-tuning path is verified in real time. If the verification passes, the fine-tuning path enters the execution stage; otherwise, it returns to be regenerated. Then, the fine-tuning path that passes the verification is converted into drive wheel control commands and sent to the corresponding AGV inbound vehicles. The specific formula for fine-tuning path verification is as follows:

[0058]

[0059] In the formula, This indicates that each AGV inbound trolley is in close contact with the first AGV during the fine-tuning process. Real-time distance between each surrounding obstacle The table shows the real-time coordinates of the AGV inbound trolley during the fine-tuning process. Indicates the first The coordinates of the obstacle.

[0060] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0061] 1. This invention acquires a static warehouse map and constructs a path map using aisle intersections, shelf gaps, and charging stations. Simultaneously, a basic path is generated between the task start point and the target location and smoothed. Real-time data collection is then performed on the position of each AGV (Automated Guided Vehicle) inbound vehicle, aisle congestion, dynamic obstacles, battery level, and task urgency to form dynamic weights. The basic path is then broken down into micro-path segments and scored. Low-scoring segments are optimized to generate a complete path. Simultaneously, path nodes are synchronized to an external central scheduling system to predict arrival times, perform conflict detection and severity assessment, and adjust the travel paths of each AGV inbound vehicle based on priority scores. Finally, shelf images are collected for coordinate transformation, and pose deviations are calculated using historical docking postures to fine-tune the poses of each AGV inbound vehicle. This improves path efficiency and robustness, enables conflict detection and priority scheduling for each AGV inbound vehicle, reduces congestion and collisions, and ensures stable and reliable inbound operations.

[0062] 2. This invention identifies and monitors the usage status of various management resources, and statistically analyzes the duration and peak periods of resource occupancy. Combined with resource reservation requests from each AGV (Automated Guided Vehicle) inbound vehicle, it assesses priorities and generates resource quota instructions, achieving efficient resource allocation and dynamic recycling. Through shelf image acquisition and pixel correction, shelf feature points are converted into three-dimensional coordinates, which are then converted into positions in the coordinate system of each AGV inbound vehicle. Simultaneously, pose deviations are calculated based on the target location's reference pose. Path constraints are set according to the performance of each AGV inbound vehicle and the surrounding environment, generating fine-tuning paths and performing real-time verification. Qualified paths are converted into drive instructions and sent to each AGV inbound vehicle, ensuring precise connection and stable execution. This guarantees the feasibility and stability of fine-tuning paths, achieving dynamic, rational, and efficient resource allocation and recycling, improving overall scheduling efficiency, and ensuring task continuity. Attached Figure Description

[0063] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0064] Figure 1 This is a system block diagram of an AGV (Automated Guided Vehicle) warehouse entry cart proposed in this invention;

[0065] Figure 2 This is a flowchart of an AGV (Automated Guided Vehicle) warehousing path planning method proposed in this invention. Detailed Implementation

[0066] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0067] Example 1

[0068] Reference Figure 1 An AGV (Automated Guided Vehicle) for warehouse entry includes a scheduling and planning module, a perception and reconstruction module, a control and navigation module, a communication and coordination module, a resource allocation module, a positioning and adjustment module, a leveling and execution module, a recovery and energy storage module, and an energy management module.

[0069] The scheduling and planning module is used to construct a dynamic weight matrix and plan the basic path for each AGV to move into the warehouse; the perception and reconstruction module is used to obtain the pose information of each AGV and the surrounding environment information in real time to dynamically generate the driving path.

[0070] Specifically, a static map of the warehouse is acquired, with aisle intersections, shelf gaps, and charging station locations as nodes, and connecting channels between nodes as edges. The starting point and target storage area in the task instruction are then obtained. Simultaneously, the actual cost from the starting point to the corresponding node and the Manhattan distance from each node to the target storage location are calculated to generate a basic path. The generated basic path is then smoothed. In real time, the current position of each AGV entering the warehouse, the congestion status of surrounding aisles, the coordinates of dynamic obstacles, the battery level of each AGV entering the warehouse, and the task urgency are collected and updated. At the same time, the parameters of the dynamic weight matrix are obtained. The basic path is divided into multiple micro-path segments at preset intervals, and the comprehensive score of each micro-path segment is calculated. Micro-path segments with scores below a preset threshold are marked as segments to be optimized. Connecting paths within a preset range around each segment to be optimized are searched to generate multiple candidate micro-path segments. The resilience value of each candidate micro-path segment is calculated and sorted from high to low. The top-ranked candidate micro-path segment is selected as the alternative path, which is then embedded into the basic path. Simultaneously, a smoothing algorithm is used to correct the connection points between the alternative path and the basic path to generate a complete path.

[0071] It should be further explained that the specific formula for calculating the actual cost is as follows:

[0072]

[0073] In the formula, Indicates the actual cost, Indicates the first in the path The weight coefficient of the segment edge, Indicates the first in the basic path The actual physical length of the segment edge.

[0074] The specific formula for calculating the Manhattan distance is as follows:

[0075]

[0076] In the formula, Indicates distance from Manhattan. Represents a node Coordinates in the warehouse plane coordinate system Indicates the target storage area The coordinates of the reference point;

[0077] The specific formula for calculating the toughness value is as follows:

[0078]

[0079] In the formula, Indicates the toughness value. Indicates the width adaptation factor. Represents the historical stability coefficient. This represents the time compatibility coefficient.

[0080] The control and navigation module is used to calculate the wheel speed and steering commands of each AGV entering the warehouse, and control each AGV to avoid obstacles in real time; the communication and coordination module is used to build the communication foundation for the coupled scheduling of task-energy-traffic flow, and generate resolution strategies.

[0081] Specifically, the system extracts each node in the complete path in real time, predicts the time for each path node, generates a corresponding estimated arrival time window, and obtains the reservation information of each AGV entering the warehouse. This information is then synchronized to the external central dispatch system and adjacent AGVs entering the warehouse within the path overlap area via a wireless network. The receiver writes the information into a local path reservation cache table and compares the path node overlap. If multiple AGVs entering the warehouse appear simultaneously within the time window of the same path node, conflict prediction is performed. Based on preset classification rules, the conflict level of the current AGV entering the warehouse is determined as high, medium, or low. The corresponding resource priority score is calculated based on the task urgency, energy criticality, path substitutability, path resilience, and historical waiting factor of each conflicting AGV entering the warehouse. These scores are then arranged from high to low. In addition, the original paths of AGVs entering the warehouse with scores higher than a preset threshold are retained according to the ranking order, while the paths of other AGVs entering the warehouse are adjusted.

[0082] The resource allocation module is used to dynamically allocate resource usage rights and time quotas based on the reservation requests sent by each AGV inbound vehicle; the positioning adjustment module is used to acquire the visual features of the shelf and calculate the pose deviation between each AGV inbound vehicle and the target storage location.

[0083] Specifically, image data of the front-end structure of the shelving is collected, and coordinate information is obtained based on the image data of each shelving front-end structure. Then, the image coordinates are converted into spatial poses in the coordinate system of each AGV inbound vehicle through calibration parameters to output the corresponding three-dimensional coarse positioning information. The docking posture of each AGV relative to the cargo location is obtained in history and compared with the self-pose of each AGV inbound vehicle and the pose of the cargo location in the three-dimensional coarse positioning information. The pose deviation between each AGV inbound vehicle and the target cargo location is calculated. The calculated pose deviation is compared with the target pose of the cargo location to calculate the compensation adjustment amount required for each AGV inbound vehicle. The pose of each AGV inbound vehicle is then finely corrected based on the calculated compensation adjustment amount.

[0084] Furthermore, this implementation requires further explanation of the specific formula for calculating the pose deviation:

[0085] The specific formula for calculating position deviation is as follows:

[0086]

[0087] In the formula, This indicates the positional deviation. Indicates the actual position of the AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, Indicates the actual position of each AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate;

[0088] The specific formula for calculating attitude angle deviation is as follows:

[0089]

[0090] In the formula, Indicates the attitude angle deviation. This indicates the actual attitude angle of each AGV entering the warehouse. This represents the expected attitude angle of each AGV entering the warehouse;

[0091] The specific formula for calculating the compensation adjustment amount is as follows:

[0092]

[0093] In the formula, This represents the pose compensation adjustment amount, which includes three components, corresponding to X-axis position compensation, Y-axis position compensation, and attitude angle compensation, respectively. This indicates the X and Y axis position compensation ratio coefficients. This represents the attitude angle compensation ratio coefficient. The X and Y coordinates represent the orientation of the target cargo location. The current actual X and Y coordinates of the AGV inbound vehicle. Target attitude angle of the cargo location The current actual attitude angle of each AGV entering the warehouse.

[0094] The leveling execution module is used to control each AGV warehouse entry trolley to perform dynamic leveling and execute pallet storage and retrieval operations; the energy recovery and storage module is used to convert waste heat into electrical energy and store the generated electrical energy; the energy management module is used to preheat or assist in cooling the battery packs of each AGV warehouse entry trolley.

[0095] Example 2

[0096] Reference Figure 2 A method for planning the inbound path of an AGV (Automated Guided Vehicle) inbound vehicle is described, and the specific steps of the planning method are as follows:

[0097] Collect and parse the inbound task information, and generate the basic path for each AGV inbound vehicle based on the static warehouse map.

[0098] The system collects the location and surrounding environment information of each AGV entering the warehouse in real time, and evaluates and adjusts the generated basic path.

[0099] Collect the path reservation information sent by each AGV inbound vehicle, predict conflicts based on the path reservation information, and coordinate various conflicts.

[0100] The system collects real-time data on resource occupancy and resource usage requests and reservations from each AGV (Automated Guided Vehicle) for inbound transport, and then dynamically allocates quotas.

[0101] Specifically, the system collects and assigns corresponding identifiers to various management resources, detects in real time whether AGVs are occupying resources within the management resource area, retrieves the occupancy records of various resources within one hour from the local resource database, and calculates the average occupancy time and peak usage periods. It collects resource reservation requests from each AGV and arranges them in ascending order, marking each request as pending. Then, it evaluates the priority of each resource reservation request and arranges them in descending order. Based on the occupancy status of various resources, it prioritizes each resource reservation request, allocates usage quotas to various resources, generates corresponding quota instructions, and sends these instructions to the corresponding AGVs. If an AGV with an allocated quota malfunctions, stops, or has its task canceled, the originally allocated quota is reclaimed and reassigned to the AGV with the highest priority in the resource reservation request.

[0102] Calculate the pose deviation between each AGV inbound trolley and the target storage location to generate the corresponding fine-tuning path.

[0103] Specifically, the system collects the reference pose data of the target storage location and the image feature points of the shelf, extracts the pixel coordinates of the shelf image feature points, corrects the pixel coordinates of the shelf image feature points using distortion coefficients, and then converts the corrected pixel coordinates into coordinates in a three-dimensional coordinate system using perspective projection formulas. These three-dimensional coordinates are then converted into coordinates in the AGV storage vehicle coordinate system to obtain the current planar position of each AGV upon unloading. The system also calculates the pose deviation between each AGV and the target storage location. Simultaneously, based on the physical performance of each AGV and the environment of the target storage location, constraint parameters for fine-tuning the path are set. Based on the calculated pose deviation and the set constraint parameters, a corresponding fine-tuning path is generated, and each fine-tuning path is verified in real time. If the verification passes, the fine-tuning path enters the execution phase; otherwise, it is regenerated. Finally, the verified fine-tuning path is converted into drive wheel control commands and sent to the corresponding AGVs.

[0104] It should be further explained that the specific formula for fine-tuning the path verification is as follows:

[0105]

[0106] In the formula, This indicates that each AGV inbound trolley is in close contact with the first AGV during the fine-tuning process. Real-time distance between each surrounding obstacle The table shows the real-time coordinates of the AGV inbound trolley during the fine-tuning process. Indicates the first The coordinates of the obstacle.

Claims

1. An AGV (Automated Guided Vehicle) for warehouse entry, characterized in that, It includes a scheduling and planning module, a perception and reconstruction module, a control and navigation module, a communication and coordination module, a resource allocation module, a positioning and adjustment module, a leveling and execution module, a recovery and storage module, and an energy management module; The scheduling and planning module is used to construct a dynamic weight matrix and plan the basic path for each AGV to move into the warehouse. The perception reconstruction module is used to acquire the pose information of each AGV entering the warehouse and the surrounding environment information in real time, so as to dynamically generate the driving path. The control and navigation module is used to calculate the wheel speed and steering commands of each AGV inbound vehicle and control each AGV inbound vehicle to avoid obstacles in real time. The communication coordination module is used to build the communication foundation for the coupled scheduling of task-energy-traffic flow and generate resolution strategies; The resource allocation module is used to dynamically allocate resource usage permissions and time quotas based on the reservation requests sent by each AGV inbound vehicle. The positioning adjustment module is used to acquire the visual features of the shelf and calculate the pose deviation between each AGV inbound trolley and the target storage location. The leveling execution module is used to control each AGV inbound trolley to perform dynamic leveling and to perform pallet storage and retrieval operations. The energy recovery and storage module is used to convert waste heat into electrical energy and store the generated electrical energy; The energy management module is used to preheat or assist in cooling the battery packs of each AGV trolley entering the warehouse. The specific steps of the perception reconstruction module in dynamically generating the driving path are as follows: S1.1: Obtain a static map of the warehouse, using aisle intersections, shelf gaps, and charging station locations as nodes, and connecting aisles between nodes as edges. Then, obtain the starting point and target storage area from the task instructions. Simultaneously, calculate the actual cost from the starting point to the corresponding node and the Manhattan distance from each node to the target storage location to generate a basic path. The specific formula for calculating the actual cost is as follows: In the formula, Indicates the actual cost, Indicates the first in the path The weight coefficient of the segment edge, Indicates the first in the basic path The actual physical length of the segment edge; The specific formula for calculating the Manhattan distance is as follows: ; In the formula, Indicates distance from Manhattan. Represents a node Coordinates in the warehouse plane coordinate system Indicates the target storage area The coordinates of the reference point; S1.2: Smooth the generated basic path, and then collect the current position of each AGV inbound vehicle, the congestion of the surrounding channels, the coordinates of dynamic obstacles, the battery level of each AGV inbound vehicle, and the task urgency update in real time, while obtaining the parameters of the dynamic weight matrix. S1.3: Divide the basic path into multiple micro-path segments according to a preset interval, calculate the comprehensive score of each micro-path segment, mark each micro-path segment with a score lower than a preset threshold as a segment to be optimized, and search for each connected path within a preset range around each segment to be optimized to generate multiple candidate micro-path segments. S1.4: Calculate the resilience value of each candidate micropath segment, rank them from highest to lowest, select the top-ranked candidate micropath segment as the alternative path, embed the alternative path into the base path, and simultaneously use a smoothing algorithm to correct the connection points between the alternative path and the base path to generate a complete path. The specific formula for calculating the resilience value is as follows: ; In the formula, Indicates the toughness value. Indicates the width adaptation factor. Represents the historical stability coefficient. This represents the time compatibility coefficient.

2. The AGV warehouse entry cart according to claim 1, characterized in that, The specific steps for the communication coordination module to generate the resolution strategy are as follows: S2.1: Extract each node in the complete path in real time, predict the time of each path node, generate the corresponding expected arrival time window, obtain the reservation information of each AGV inbound vehicle, and synchronize it to the external central dispatch system and adjacent AGV inbound vehicles in the path overlap area via wireless network. S2.2: The receiver writes the information into the local path reservation cache table, compares the path node overlap, and if multiple AGV inbound vehicles appear at the same time window on the same path node, then conflict prediction is performed, and the conflict level of the current AGV inbound vehicle is judged according to the preset division rules as high level, medium level and low level. S2.3: Calculate the corresponding resource priority score based on the task urgency, energy criticality, path substitutability, path resilience, and historical waiting factor of each conflicting AGV inbound vehicle, and sort them from high to low. At the same time, retain the original path of the AGV inbound vehicle with a score higher than the preset threshold according to the ranking order, and adjust the path of other AGV inbound vehicles.

3. The AGV warehouse entry cart according to claim 2, characterized in that, The specific steps for the positioning adjustment module to calculate the pose deviation between each AGV inbound trolley and the target cargo location are as follows: S3.1: Collect image data of the front structure of the shelf, obtain coordinate information based on the image data of the front structure of each shelf, and then convert the image coordinates into the spatial pose of each AGV inbound trolley in the coordinate system through calibration parameters, so as to output the corresponding three-dimensional coarse positioning information. S3.2: Obtain the docking posture of each AGV relative to the cargo location in history, and compare it with the self-pose of each AGV inbound vehicle and the pose of the cargo location in the 3D coarse positioning information. Calculate the pose deviation between each AGV inbound vehicle and the target cargo location. The specific formula for calculating the pose deviation is as follows: The specific formula for calculating pose deviation is as follows: ; In the formula, This indicates the pose deviation. Indicates the actual position of the AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, This indicates the actual position of each AGV (Automated Guided Vehicle) trolley entering the warehouse. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate, This indicates the desired position of each AGV (Automated Guided Vehicle) inbound trolley. coordinate; The specific formula for calculating attitude angle deviation is as follows: ； In the formula, Indicates the attitude angle deviation. This indicates the actual attitude angle of each AGV entering the warehouse. This represents the expected attitude angle of each AGV entering the warehouse; S3.3: Compare the calculated pose deviation with the target pose of the storage location, calculate the compensation adjustment required for each AGV inbound trolley, and make minor corrections to the pose of each AGV inbound trolley based on the calculated compensation adjustment. The specific formula for calculating the compensation adjustment is as follows: ; In the formula, This represents the pose compensation adjustment amount, which includes three components, corresponding to X-axis position compensation, Y-axis position compensation, and attitude angle compensation, respectively. Indicates the X and Y axis position compensation ratio coefficients. This represents the attitude angle compensation ratio coefficient. The X and Y coordinates represent the orientation of the target cargo location. This indicates the current actual X and Y coordinates of the AGV (Automated Guided Vehicle) for inbound operations. Indicates the target attitude angle of the cargo location. This indicates the current actual attitude angle of each AGV entering the warehouse.

4. A method for planning the inbound path of an AGV (Automated Guided Vehicle) warehouse trolley, used to implement the function of the AGV warehouse trolley as described in any one of claims 1-3, characterized in that, The specific steps of this planning method are as follows: Ⅰ: Collect and parse the inbound task information, and generate the basic path for each AGV inbound vehicle based on the static warehouse map; II: Real-time collection of the location and surrounding environment information of each AGV entering the warehouse, evaluation and adjustment of the generated basic path; Ⅲ: Collect the path reservation information sent by each AGV inbound vehicle, predict conflicts based on the path reservation information, and coordinate various conflicts. IV: Collect real-time resource occupancy data and resource usage requests and reservations from each AGV (Automated Guided Vehicle) for inbound trolleys, and dynamically allocate quotas; V: Calculate the pose deviation between each AGV inbound trolley and the target storage location to generate the corresponding fine-tuning path.

5. The AGV warehousing vehicle path planning method according to claim 4, characterized in that, The specific steps for dynamic quota allocation described in step IV are as follows: S4.1: Collect and assign corresponding identifiers to various management resources, detect in real time whether there are AGV inbound vehicles occupying resources in the management resource area, and retrieve the occupancy records of various resources within 1 hour from the local resource database, and calculate the average occupancy time and peak usage period. S4.2: Collect resource reservation requests from each AGV inbound vehicle, arrange each resource reservation request in ascending order, mark each resource reservation request as pending, evaluate the priority of each resource reservation request, and arrange them in descending order. S4.3: Based on the resource occupancy status, prioritize each resource reservation request, allocate usage quotas for each type of resource, generate corresponding quota instructions, and then send the quota instructions to the corresponding AGV inbound vehicle. If the AGV inbound vehicle with the allocated quota malfunctions or the task is canceled, the originally allocated quota is reclaimed and reassigned to the AGV inbound vehicle with the highest priority of the resource reservation request.

6. The AGV warehousing vehicle path planning method according to claim 5, characterized in that, The specific steps for generating the corresponding fine-tuning path in step V are as follows: S5.1: Collect the reference pose data of the target location and the image feature points of the shelf, extract the pixel coordinates of the shelf image feature points, correct the pixel coordinates of the shelf image feature points through the distortion coefficient, and then convert the corrected pixel coordinates into coordinates in the three-dimensional coordinate system through the perspective projection formula. S5.2: Convert the coordinates in the three-dimensional coordinate system to the coordinates in the AGV loading vehicle coordinate system to obtain the current planar position of each AGV loading vehicle and calculate the pose deviation between each AGV loading vehicle and the target storage location. At the same time, set the constraint parameters for fine-tuning the path based on the physical performance of each AGV loading vehicle and the environment of the target storage location. S5.3: Based on the calculated pose deviation and the constraint parameters of the set fine-tuning path, a corresponding fine-tuning path is generated, and each fine-tuning path is verified in real time. If the verification passes, the fine-tuning path enters the execution stage; otherwise, it returns to be regenerated. Then, the fine-tuning path that passes the verification is converted into drive wheel control commands and sent to the corresponding AGV inbound vehicles. The specific formula for fine-tuning path verification is as follows: ; In the formula, This indicates that each AGV inbound trolley is in close contact with the first AGV during the fine-tuning process. Real-time distance between each surrounding obstacle The table shows the real-time coordinates of the AGV inbound trolley during the fine-tuning process. Indicates the first The coordinates of the obstacle.

Images