Automatic loading and unloading method and system based on AGV

Abstract

The invention discloses an AGV-based automatic loading and unloading method and system, and relates to the technical field of industrial vehicle automatic driving, and the unloading process comprises the steps that after a loading and unloading vehicle arrives at a parking area, an AGV forklift drives to a designated path according to a station path issued by an AGV scheduling system, and an unloading task issued by the AGV scheduling system is obtained; the AGV forklift collects an image of an area where a to-be-forked carrier is located according to a depth camera, calculates carrier pose information, generates a dynamic path in combination with position coordinates of the AGV forklift, and carries out tail end unloading operation according to the dynamic path until unloading is completed; the loading process comprises the steps that after a loading and unloading vehicle arrives at a goods taking and placing channel area, a radar pan-tilt scans the position and posture of a compartment of the loading and unloading vehicle and dynamically calculates the positions and poses of multiple storage positions in the compartment, serial numbers and the positions of corresponding storage position front points; after the AGV forklift travels to a specified storage location preposition point according to a loading task issued by the AGV dispatching system and the current loading storage location position issued by the radar holder, the AGV forklift executes a precise loading task at the loading preposition point through the tail end recognition system, and tail end loading operation is conducted according to a dynamic path of the AGV forklift until loading of all storage locations is completed; according to the automatic loading and unloading method and system, the operation efficiency and the automation level of a logistics system are remarkably improved, and reliable flexible logistics guarantee is provided for intelligent manufacturing.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology for industrial vehicles, and in particular to an AGV-based automatic loading and unloading method and system. Background Technology

[0002] Driven by the wave of digital transformation and smart factory construction, full-process automation from warehouse management and production line logistics to last-mile delivery has become an industry trend, and automated loading and unloading, as an indispensable part of this process, is ushering in a period of rapid development.

[0003] Currently, the technologies for bagged material loading robots and automated pallet loading systems are relatively mature and have been applied in real-world scenarios. However, the market demand for boxed goods loading robots is the strongest, but due to their high technical complexity (such as multi-category box recognition, dynamic grasping planning, and precise stacking), a fully mature solution has not yet emerged, and there is still considerable room for technological breakthroughs.

[0004] Loading and unloading operations have always been labor-intensive, involving heavy manual labor and high labor costs, making intelligent transformation urgently needed. However, development on the supply side remains lagging. Despite rising market enthusiasm, only a few companies possess truly mature products, and the industry as a whole is still in its early stages. There is an urgent need for collaboration across the entire industry chain to jointly build a healthy and sustainable industrial ecosystem and promote the large-scale implementation of automated loading and unloading technology.

[0005] Existing AGV-based automated material handling systems are not compatible with the site conditions, the AGV guidance paths are not optimized, and the system maintenance costs are high, which hinders the application of AGVs. Summary of the Invention

[0006] Based on the technical problems existing in the background technology, this invention proposes an AGV-based automatic loading and unloading method and system, which significantly improves the operating efficiency and automation level of the logistics system and provides reliable flexible logistics support for intelligent manufacturing.

[0007] The present invention proposes an AGV-based automatic loading and unloading method, comprising: When the loading and unloading vehicle arrives at the parking area, the AGV forklift travels to the designated path according to the station path issued by the AGV scheduling system, and then obtains the unloading task issued by the AGV scheduling system. The AGV scheduling system plans the station path according to the position and quantity of the loading and unloading vehicle issued by the radar pan-tilt unit. The AGV forklift uses a depth camera to capture images of the area where the vehicle to be picked up is located, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs the final unloading operation according to the dynamic path until unloading is complete.

[0008] Furthermore, the AGV scheduling system plans the station path based on the pose of the loading / unloading vehicle and the quantity of goods issued by the radar pan-tilt unit, specifically as follows: The radar pan-tilt unit performs a full-area scan to obtain raw data, and then converts the raw data into a map coordinate system based on the calibration extrinsic parameters. The loading area of ​​the loading and unloading vehicle is extracted from the transformed data using a vehicle compartment extraction algorithm. The storage location planning algorithm automatically plans the storage location based on the loading area, the size of the goods to be loaded, and the placement gap requirements. The AGV scheduling system uses the location of the storage space as a coarse target, plans the path for the AGV forklift to travel to the coarse target, and sends this path as the station path to the AGV forklift.

[0009] Furthermore, the vehicle's pose information is calculated based on the AGV end-effector identification system, specifically as follows: When the AGV forklift travels to the unloading point, the depth camera collects RGB images and depth data of the area where the vehicle to be picked up is located; Identify vehicles in RGB images and segment the images; Calculate the pixel coordinates of the center point of the support column region in the RGB image. Based on the pixel coordinates of the center point and the depth information, calculate the relative height of the vehicle relative to the horizontal plane of the camera's optical center. and the position (horizontal offset) of the vehicle in the camera coordinate system and vertical offset ); Calculate the height of the fixture socket center relative to the fork plane in a multi-layer scene by using the relative height between the forks and the camera. Based on the real-time height of the AGV forklift fork tip Calculate the height at which the forks need to be raised or lowered. Based on this, the vehicle's pose information can be obtained; in, , A positive value indicates that a certain lifting height is required. A negative value indicates that a descent is required.

[0010] Furthermore, the AGV end-of-line identification system includes an end-of-line identification controller, a depth camera, and calibration tools; The end-point recognition controller is connected to the depth camera via a network cable. The depth camera is calibrated using a calibration tool before use. The end-point identification controller is connected to the switch via Ethernet, forming a local area network with the vehicle controller and navigation device; The end-point identification controller communicates directly with the vehicle controller via the CAN bus.

[0011] Furthermore, the process for generating the position coordinates of the AGV forklift is as follows: Reflective barrels are set up, and the AGV forklift uses the navigation head to detect the reflective barrels within its field of vision in real time and measure the distance between the AGV forklift and each reflective barrel; Based on the measured distance and the known coordinates of the reflectors, the AGV scheduling system calculates the position coordinates of the AGV forklift in real time.

[0012] Furthermore, in conventional unloading operations, the AGV forklift travels along a pre-set fixed path; During the task execution, the AGV forklift is guided to the fixed work point based on real-time positioning data to complete the corresponding operation.

[0013] An AGV-based automated loading method includes: When the loading and unloading vehicle arrives at the picking and placing channel area, the AGV forklift drives to the designated storage location according to the storage location issued by the AGV scheduling system, and then obtains the loading task issued by the AGV scheduling system. The AGV scheduling system calculates the division position of each storage location on the loading platform based on the size information of the loading platform obtained by the radar pan-tilt unit scanning the loading area and the size of the goods to be loaded. AGV forklifts perform loading tasks in the picking lane through the end-of-line identification system. Based on multiple storage location numbers generated by the system, for each storage location, the AGV forklifts collect images of the area where the storage location is located using a depth camera, calculate loading and unloading feature information, and generate a dynamic path by combining the AGV forklifts' position coordinates. The AGV forklifts then perform end-of-line loading operations according to the dynamic path until all storage locations are loaded.

[0014] Furthermore, in a typical pickup operation scenario, the AGV forklift travels along a pre-set fixed path; During the task execution, the AGV forklift is guided to the fixed work point based on real-time positioning data to complete the corresponding operation.

[0015] An AGV-based automated unloading system includes a coarse positioning module and a fine positioning module for unloading. In the coarse positioning module for unloading, after the loading and unloading vehicle arrives at the parking area, the AGV forklift travels to the designated path according to the station path issued by the AGV scheduling system, and then obtains the unloading task issued by the AGV scheduling system. The AGV scheduling system plans the station path according to the pose of the loading and unloading vehicle and the quantity of goods issued by the radar pan-tilt unit. In the unloading precision positioning module, the AGV forklift uses a depth camera to capture images of the area where the vehicle to be picked up is located, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs the end-of-line unloading operation according to the dynamic path until unloading is complete.

[0016] An AGV-based automated loading system includes a coarse positioning module and a fine positioning module for loading. In the loading coarse positioning module, after the loading and unloading vehicle arrives at the picking and placing channel area, the AGV forklift drives to the designated storage location according to the storage location issued by the AGV scheduling system, and then obtains the loading task issued by the AGV scheduling system. The AGV scheduling system calculates the division position of each storage location on the loading platform based on the size information of the loading platform obtained by the radar pan-tilt unit scanning the loading area and the size of the goods to be loaded. In the loading and positioning module, the AGV forklift performs loading tasks in the picking channel through the end-of-line identification system. Based on the multiple storage location numbers generated by the system, for each storage location, the AGV forklift collects images of the area where the storage location is located using a depth camera, calculates loading and unloading feature information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs end-of-line loading operations according to the dynamic path until all storage locations are loaded.

[0017] The advantages of the AGV-based automated loading and unloading method and system provided by this invention are: it is an intelligent solution based on the AGV platform, deeply integrating advanced technologies such as point cloud computing, computer vision, and edge computing. This system is mainly applied to key operational scenarios such as AGV docking positioning, precise docking, and automated loading and unloading. Through intelligent sensing technology, it identifies the precise pose and status information of end targets such as pallets, shelves, cages, steel drums, and material bags in real time. The system employs high-precision algorithms, achieving automated loading and unloading operations with millimeter-level accuracy, significantly improving the operational efficiency and automation level of the logistics system, and providing reliable flexible logistics support for intelligent manufacturing. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a schematic diagram illustrating the structural connection between the end-point identification system and the vehicle-mounted device. Figure 3 A flowchart for path planning in an end-point identification system; Figure 4 This is a schematic diagram of the AGV forklift's path. Figure 5 A flowchart for the loading and unloading process. Detailed Implementation

[0019] The technical solution of the present invention will now be described in detail through specific embodiments. Many specific details are set forth in the following description to provide a thorough understanding of the invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0020] Against the backdrop of the rapid development of intelligent manufacturing and smart logistics, AGVs (Automated Guided Vehicles), as core equipment of flexible logistics systems, are accelerating their evolution towards intelligence and autonomy. Among them, the AGV automated loading and unloading system, as a key technological breakthrough, significantly improves the end-point operation capability of AGVs in complex working conditions through intelligent sensing, dynamic path planning, and precise control, achieving efficient, stable, and accurate automated loading and unloading operations.

[0021] Firstly, this embodiment can realize automatic loading and unloading, mainly relying on the automatic loading and unloading system. The automatic loading and unloading system consists of a radar gimbal, AGV forklifts, and end recognition system. It is divided into three areas: the loading and unloading parking area, the loading and unloading channel area, and the picking and placing channel area.

[0022] like Figures 1 to 5 As shown, the present invention proposes an AGV-based automatic unloading method, comprising: S10. When the loading and unloading vehicle arrives at the parking area, the AGV forklift travels to the designated path according to the station path issued by the AGV scheduling system, and then obtains the unloading task issued by the AGV scheduling system. The AGV scheduling system plans the station path according to the position and quantity of the loading and unloading vehicle issued by the radar pan-tilt unit. S20. The AGV forklift acquires images of the area where the vehicle to be picked up is located using a depth camera, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs the end-of-line unloading operation according to the dynamic path until unloading is complete.

[0023] In addition, this embodiment also proposes an AGV-based automatic loading method, characterized by including: S01. When the loading and unloading vehicle arrives at the picking and placing channel area, the AGV forklift drives to the designated storage location according to the storage location issued by the AGV scheduling system, and then obtains the picking task issued by the AGV scheduling system. The AGV scheduling system calculates the division position of each storage location on the loading platform based on the size information of the loading platform obtained by the radar pan-tilt unit scanning the loading area and the size of the goods to be loaded. S02. The AGV forklift performs loading tasks in the picking channel through the end recognition system. Based on the multiple storage location numbers generated by the system, for each storage location, the AGV forklift collects images of the area where the storage location is located using a depth camera, calculates loading and unloading feature information, and generates a dynamic path by combining the position coordinates of the AGV forklift. The AGV forklift performs end loading operations according to the dynamic path until all storage locations are loaded. That is, the radar gimbal scans the position of the loading and unloading vehicle compartment and dynamically calculates the position, sequence number and corresponding front point of multiple storage locations in the compartment. The AGV forklift travels to the designated front point of the storage location according to the loading task issued by the AGV scheduling system and the current loading storage location issued by the radar gimbal. At the loading front point, the AGV forklift performs a precise loading task through the end recognition system and performs end loading operation according to its dynamic path until all storage locations are loaded.

[0024] This embodiment presents an intelligent solution based on an AGV platform, deeply integrating advanced technologies such as point cloud computing, computer vision, and edge computing. The system is primarily applied to key operational scenarios such as AGV docking and positioning, precise docking, and automated loading and unloading. Through intelligent sensing technology, it identifies the precise pose and status information of end-point targets such as pallets, shelves, cages, drums, and bags in real time. Employing high-precision algorithms, the system achieves automated loading and unloading operations with millimeter-level accuracy, significantly improving the operational efficiency and automation level of the logistics system, and providing reliable flexible logistics support for intelligent manufacturing.

[0025] In this embodiment, the overall logic of the unloading method and the loading method is similar, and they will be described together below.

[0026] (a1) When the loading / unloading vehicle arrives at the parking area, the radar pan-tilt unit obtains the vehicle's position and the quantity of goods through point cloud computing and sends this information to the AGV scheduling system. The AGV scheduling system plans the station path based on the radar pan-tilt unit data and completes coarse positioning. After the AGV forklift travels to the designated path, it issues an unloading task to the end-of-line identification system.

[0027] like Figure 5 As shown, the radar gimbal performs coarse positioning of the loading / unloading vehicle and the quantity of goods. The execution process is as follows: Step 1.1: The WMS system (Warehouse Management System) sends a signal to the controller to begin loading; Step 1.2: The controller sends a scanning signal to the radar pan-tilt unit to start full-area scanning. The radar pan-tilt unit has a scanning sector of 0-180° in three-dimensional space. After the scanning is completed, the raw scanning data is sent to the controller. Step 1.3: The controller converts the raw data from the scanning device in Step 1.2 into the existing map coordinate system based on the calibration extrinsic parameters; Step 1.4: The compartment extraction algorithm uses the data transformed in Step 1.3 to extract the loadable area of ​​the loading and unloading vehicle.

[0028] Step 1.5: The storage location planning algorithm system uses the loadable area calculated in Step 1.4 and automatically plans the storage location based on the configured parameters such as the size of the goods to be loaded and the placement gap requirements.

[0029] Step 1.6: Based on the warehouse location calculated in Step 1.5, automatically plan a path that can be driven to the warehouse location, and send the path to the AGV forklift through the calculation device.

[0030] Step 1.7: The AGV receives the work instructions sent by the scheduling system and completes the path acquisition and repositioning (coarse positioning) of the AGV according to the new path.

[0031] It should be noted that the core objective of the compartment extraction algorithm is to accurately identify the available loading area inside the compartment of a loading and unloading vehicle (usually a truck or container) from radar point cloud data and separate it from the surrounding environment (such as the cab, supports, ground, etc.).

[0032] The specific steps of the carriage extraction algorithm are as follows: (b1) Data preprocessing; Filtering and noise reduction: Filter the original radar point cloud to remove discrete noise points caused by dust, smoke or the sensor itself.

[0033] Ground segmentation: Considering the significant difference in depth values ​​between the carriage point cloud data and the ground point cloud data, the carriage data is located by considering the depth data values ​​of the ground point cloud.

[0034] (b2) Point cloud clustering: After filtering out the ground point cloud data, clustering algorithms (such as Euclidean clustering and DBSCAN) are used to segment it into different "object blocks," distinguishing the truck bed, cab, and adjacent pillars.

[0035] (b3) Carriage identification and extraction: Feature recognition: After edge extraction algorithms, identify a large cuboid point cloud cluster with specific geometric features. These features include: (c1) Planar structure: The bottom, side walls and front end (the end closest to the front of the car) of the carriage are usually large planes.

[0036] (c2) Size constraints: The system will set an approximate size range (length, width, height) for the expected carriage. Objects that do not conform to this range (such as trolleys and pedestrians) will be excluded.

[0037] (c3) Spatial relationship: The carriage is usually located in a specific position in the scanning area, and its bottom is slightly higher than the ground (because of the tires and chassis).

[0038] (c4) Bounding box fitting: Once the point cloud cluster that is most likely the carriage is identified, the algorithm will fit it with a 3D bounding box. This bounding box represents the physical space of the entire carriage.

[0039] (c4) Loadable region extraction: This is the ultimate goal. The algorithm will calculate the interior space of this 3D bounding box.

[0040] Specifically, the loadable area extraction process determines: the position and planar equations of the carriage floor, and the positions of the four inner walls of the carriage. This results in a clear, unobstructed three-dimensional spatial area suitable for placing goods.

[0041] It should be noted that the core objective of the storage location planning algorithm is to automatically and efficiently calculate the precise location and orientation of each item within a known "loadable area" based on the size and clearance requirements of the goods to be loaded, i.e., the "storage location".

[0042] The steps of the storage location planning algorithm: (d1) Input parameters: Loadable area: The output of the "carriage extraction algorithm" is a three-dimensional space; Cargo dimensions: length, width, and height of the pallet or container; Clearance requirements: A safety clearance (e.g., 50mm) must be reserved between goods and between goods and the truck bed walls to ensure that no collision occurs during forklift operation; Placement rules (optional): For example, depending on whether the goods must be placed parallel to the carriage wall (i.e., cannot be rotated), or whether there are special stability requirements (such as from the inside out, from the center to the sides, etc.).

[0043] (d2) Spatial discretization of storage locations: The algorithm projects the three-dimensional "loadable area" onto the base plate as a two-dimensional grid map; the size of each grid is determined by the cargo size and clearance.

[0044] (d3) Storage location generation and sorting: Candidate position generation: The algorithm starts from the innermost part of the carriage, usually a corner (such as the left or right rear corner), and attempts to place the first item. After placement, the next available empty position is calculated based on the gap requirements.

[0045] Placement strategy: This is the core logic of the algorithm. Common strategies include: Row placement: Place the items sequentially along the width (lateral) of the carriage. After one row is filled, start a new row towards the carriage door. Column placement: Place them sequentially along the length of the carriage (longitudinal direction). After one column is full, move them laterally to start a new column. Block placement: Suitable for handling multiple goods at once (such as picking up two pallets at once), treating multiple goods as a whole "block" for placement; Optimization goals: The algorithm will primarily aim to maximize space utilization and optimize placement stability. For example, it may prioritize leaving gaps on one side of the carriage door rather than spreading them throughout the interior; or ensure that the placement pattern is symmetrical to maintain vehicle balance.

[0046] (d4) Output results: The algorithm ultimately outputs a list of storage locations. Each storage location in the list contains the coordinates of its center point in the large map coordinate system. and orientation angle ; The AGV scheduling system uses these specific warehouse locations as precise positioning target points for the AGV forklifts.

[0047] In summary, the carriage extraction algorithm → "environment perception": solves the problem of "where the carriage is and how big it is". It uses radar point cloud data to reconstruct the geometric model of the carriage; Storage location planning algorithm → "Decision Planning": solves the problem of "how should goods be placed". Based on known carriage space and cargo rules, it calculates the optimal placement layout; The close integration of the cargo compartment extraction algorithm and the storage location planning algorithm enables the AGV scheduling system not only to "see" the vehicle, but also to "think" about how to neatly and efficiently load or unload the goods. This is a key technology for achieving fully automated loading and unloading.

[0048] Understandably, during loading, the radar gimbal uses lidar data to construct a three-dimensional digital twin model of the loading vehicle space in real time, and develops a dynamic storage location allocation algorithm for space optimization based on the physical size characteristics of the cargo and the truck flatbed.

[0049] (a2) The AGV forklift collects images of the area where the vehicle to be picked up is located using a depth camera, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The end-point recognition system includes an end-point recognition controller, a depth camera, and calibration tools. The controller and camera are connected via a network cable. The camera needs to be calibrated using the calibration tools (installed on a PC) before use. The end-point recognition controller connects to a switch via Ethernet, forming a local area network with the vehicle controller and navigation device. The end-point recognition controller can also be directly connected to the CAN bus to communicate directly with the vehicle controller.

[0050] Camera calibration software algorithm: A low-cost depth camera is used as a sensor, mounted at the root of the fork. Using the vehicle center as a reference, the extrinsic parameters of the depth camera are calibrated, including the horizontal offset (dx), vertical offset (dy), and height offset (dz, or h) relative to the vehicle center, as well as pitch, roll, and yaw angles. The camera extrinsic parameters are expressed as the transformation relationship between the camera coordinates and the vehicle center coordinate system, i.e., solving the transformation between the two three-dimensional coordinate systems into rigid body motion. Alternatively, the extrinsic parameters can be represented by a rotation matrix R and a translation matrix T. Specific calibration calculation methods and transformation relationships can be achieved using existing technologies, which will not be elaborated in this embodiment.

[0051] When the AGV forklift travels to the pre-position of the end-of-line operation (picking or loading), the depth camera acquires RGB images and depth data of the area where the vehicle to be picked up is located; the vehicle in the RGB image is identified and the image is segmented; the pixel coordinates of the center point of the support column area in the RGB image are calculated; based on the center point pixel coordinates and depth information, the relative height of the vehicle relative to the horizontal plane of the camera's optical center is calculated; through the relative height of the forks and the camera, the height of the center of the vehicle's insertion hole relative to the ground in multi-level scenes is accurately calculated; based on the real-time height of the AGV forklift fork tip, the height at which the forks need to be raised or lowered is calculated, thereby adjusting the fork tip height to pick up goods during the picking process, realizing the positioning of the end-of-line operation object through multi-fusion processing in complex scenes, forming a unified attitude calculation and local path planning method, and controlling the vehicle's movement to complete the complex end-of-line operation process.

[0052] During the deployment phase of the AGV scheduling system, the precise coordinates of all reflectors are acquired. The AGV vehicle uses its navigation head to detect the reflectors within its field of view in real time and accurately measures the distance to each reflector. Based on this distance data and the known coordinates of the reflectors, the system can calculate the precise position coordinates of the AGV vehicle in real time.

[0053] like Figure 4 As shown, in a typical loading / unloading / placing operation scenario, the AGV forklift travels along a pre-set fixed path. During task execution, the system guides the AGV forklift precisely to the fixed work point based on real-time positioning data to complete the corresponding operation.

[0054] In the loading and unloading area, the system employs a dynamic path planning strategy. When an AGV forklift enters this area, it proactively sends a path request to the gimbal system and the end-of-line identification system. The gimbal system monitors the truck's position in real time, and the end-of-line identification system accurately tracks the cargo status. Together, they generate the optimal dynamic path and send it to the AGV forklift. The AGV forklift flexibly adjusts its route based on real-time instructions, achieving efficient and accurate picking and placing of goods in non-fixed locations. This dynamic path planning mechanism significantly enhances the system's adaptability and operational flexibility under complex working conditions.

[0055] It should be noted that the loading and unloading steps are exactly the reverse. The radar gimbal first scans the loading area, performs pose recognition, and obtains the size information of the loading and unloading platform. Based on the size of the goods to be loaded, it calculates the division of each storage location on the loading platform (i.e., the pose, sequence number, and corresponding front point position of multiple storage locations in the truck). The AGV forklift performs the loading task in the picking lane through the end-of-line recognition system, and according to the multiple storage location sequence numbers generated by the system, the AGV forklift and its end-of-line recognition device accurately place the goods into the corresponding storage locations in sequence.

[0056] Based on the physical dimensions of the goods and loading platform, the gimbal coarse positioning algorithm calculates the truck's storage location. After obtaining the coarsely positioned storage location, the AGV system constructs dynamic storage location coordinates. Compared with the unloading step, the main difference lies in the end-point recognition stage, where loading and unloading feature information is identified to complete fine positioning.

[0057] In this process, for each truck's storage location, the AGV scheduling system and the end-of-line operation system work together to complete the loading. First, the end-of-line identification system executes the pickup / loading task at the delivery point. When performing the pickup task, the system accurately retrieves the goods from the pickup area according to the end-of-line identification and local path planning method in step S20. When performing the loading task, the system places the goods into the corresponding storage location according to the end-of-line identification and local path planning method in step S02. The loading process is as follows: Figure 5 When the AGV forklift arrives at the dynamic storage location forward point issued by the AGV scheduling system, the end-point identification system performs high-precision positioning and local path planning according to the method in step S02.

[0058] The above loading and unloading methods can achieve the following core functions: Handling complex scenarios: Improve the autonomy and environmental adaptability of AGVs to meet the needs of application scenarios such as offset, stacking, settling, multi-layer, and tilting, and improve the accuracy, smoothness and fault tolerance of operations; LiDAR point cloud computing: LiDAR is used to acquire point cloud data in real time, and the attitude of the loading and unloading vehicle target is estimated through point cloud computing. Depth camera recognition and localization: Employing RGBD multimodal perception and recognition technology, it supports real-time detection and precise localization of multiple types of targets, with errors controlled within ±10mm; Edge intelligent processing: Adopting an edge computing architecture, it realizes local recognition and real-time feedback without relying on cloud servers, thus improving system response speed; Local path planning: Construct local path planning algorithms for AGV vehicle body, vehicle pose, and loading / unloading vehicles to complete functions such as automatic loading / unloading and picking up / placing goods.

[0059] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An AGV-based automated unloading method, characterized in that, include: When the loading and unloading vehicle arrives at the parking area, the AGV forklift travels to the designated path according to the station path issued by the AGV scheduling system, and then obtains the unloading task issued by the AGV scheduling system. The AGV scheduling system plans the station path according to the position and quantity of the loading and unloading vehicle issued by the radar pan-tilt unit. The AGV forklift uses a depth camera to capture images of the area where the vehicle to be picked up is located, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs the final unloading operation according to the dynamic path until unloading is complete.

2. The method according to claim 1, characterized in that, The AGV scheduling system plans station routes based on the pose of the loading / unloading vehicles and the quantity of goods issued by the radar pan-tilt unit, specifically as follows: The radar pan-tilt unit performs a full-area scan to obtain raw data, and then converts the raw data into a map coordinate system based on the calibration extrinsic parameters. The loading area of ​​the loading and unloading vehicle is extracted from the transformed data using a vehicle compartment extraction algorithm. The storage location planning algorithm automatically plans the storage location based on the loading area, the size of the goods to be loaded, and the placement gap requirements. The AGV scheduling system uses the location of the storage space as a coarse target, plans the path for the AGV forklift to travel to the coarse target, and sends this path as the station path to the AGV forklift.

3. The method according to claim 1, characterized in that, The AGV end-effector identification system calculates the vehicle's pose information, specifically: When the AGV forklift travels to the unloading point, the depth camera collects RGB images and depth data of the area where the vehicle to be picked up is located; Identify vehicles in RGB images and segment the images; Calculate the pixel coordinates of the center point of the support column region in the RGB image. Based on the pixel coordinates of the center point and the depth information, calculate the relative height of the vehicle relative to the horizontal plane of the camera's optical center. and the position of its vehicle in the camera coordinate system; Calculate the height of the fixture socket center relative to the fork plane in a multi-layer scene by using the relative height between the forks and the camera. Based on the real-time height of the AGV forklift fork tip Calculate the height at which the forks need to be raised or lowered. Based on this, the vehicle's pose information can be obtained; in, , A positive value indicates that a certain lifting height is required. A negative value indicates that a descent is required.

4. The method according to claim 1, characterized in that, The AGV end-of-line identification system includes an end-of-line identification controller, a depth camera, and calibration tools; The end-point recognition controller is connected to the depth camera via a network cable. The depth camera is calibrated using a calibration tool before use. The end-point identification controller is connected to the switch via Ethernet, forming a local area network with the vehicle controller and navigation device; The end-point identification controller communicates directly with the vehicle controller via the CAN bus.

5. The method according to claim 1, characterized in that, The process of generating the position coordinates of the AGV forklift is as follows: Reflective barrels are set up, and the AGV forklift uses the navigation head to detect the reflective barrels within its field of vision in real time and measure the distance between the AGV forklift and each reflective barrel; Based on the measured distance and the known coordinates of the reflectors, the AGV scheduling system calculates the position coordinates of the AGV forklift in real time.

6. The method according to claim 1, characterized in that, In a typical unloading operation, the AGV forklift travels along a pre-set fixed path; During the task execution, the AGV forklift is guided to the fixed work point based on real-time positioning data to complete the corresponding operation.

7. An AGV-based automated loading method, characterized in that, include: When the loading and unloading vehicle arrives at the picking and placing channel area, the AGV forklift drives to the designated storage location according to the storage location issued by the AGV scheduling system, and then obtains the loading task issued by the AGV scheduling system. The AGV scheduling system calculates the division position of each storage location on the loading platform based on the size information of the loading platform obtained by the radar pan-tilt unit scanning the loading area and the size of the goods to be loaded. AGV forklifts perform loading tasks in the picking lane through the end-of-line identification system. Based on multiple storage location numbers generated by the system, for each storage location, the AGV forklifts collect images of the area where the storage location is located using a depth camera, calculate loading and unloading feature information, and generate a dynamic path by combining the AGV forklifts' position coordinates. The AGV forklifts then perform end-of-line loading operations according to the dynamic path until all storage locations are loaded.

8. The method according to claim 7, characterized in that, In a typical picking operation scenario, the AGV forklift travels along a preset fixed path; During the task execution, the AGV forklift is guided to the fixed work point based on real-time positioning data to complete the corresponding operation.

9. An AGV-based automated unloading system, characterized in that, Includes a coarse positioning module for unloading and a fine positioning module for unloading; In the coarse positioning module for unloading, after the loading and unloading vehicle arrives at the parking area, the AGV forklift travels to the designated path according to the station path issued by the AGV scheduling system, and then obtains the unloading task issued by the AGV scheduling system. The AGV scheduling system plans the station path according to the pose of the loading and unloading vehicle and the quantity of goods issued by the radar pan-tilt unit. In the unloading precision positioning module, the AGV forklift uses a depth camera to capture images of the area where the vehicle to be picked up is located, calculates the vehicle's pose information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs the end-of-line unloading operation according to the dynamic path until unloading is complete.

10. An AGV-based automated loading system, characterized in that, Includes a coarse positioning module and a fine positioning module for loading; In the loading coarse positioning module, after the loading and unloading vehicle arrives at the picking and placing channel area, the AGV forklift drives to the designated storage location according to the storage location issued by the AGV scheduling system, and then obtains the picking task issued by the AGV scheduling system. The AGV scheduling system calculates the division position of each storage location on the loading platform based on the size information of the loading platform obtained by the radar pan-tilt unit scanning the loading area and the size of the goods to be loaded. In the loading precision positioning module, the AGV forklift performs loading tasks in the picking channel through the end-of-line identification system. Based on the multiple storage location numbers generated by the system, for each storage location, the AGV forklift collects images of the area where the storage location is located using a depth camera, calculates loading and unloading feature information, and generates a dynamic path by combining the AGV forklift's position coordinates. The AGV forklift then performs end-of-line loading operations according to the dynamic path until all storage locations are loaded.

Images