A production line unloading AGV stacker carrying control method and system

Abstract

The present application relates to the technical field of intelligent logistics automation, and particularly relates to a production line unloading AGV stacker carrying control method and system, the method comprising: constructing an environment-task-equipment multidimensional perception unit, modeling and accurately positioning the scene through laser SLAM, visual recognition and RFID positioning; dynamically planning a collision-free path based on an improved A algorithm, designing a dynamic obstacle weight factor and a production line real-time congestion index, and generating an optimal path according to the dynamic obstacle weight factor and the production line real-time congestion index; adjusting the AGV travel speed and the stacker lifting speed using a fuzzy PID algorithm, and realizing the action connection of the two through a time axis synchronization mechanism; constructing an AGV stacker carrying task priority scheduling model, and optimizing the carrying timing through digital twin pre-rehearsal. The present application improves the production line material flow efficiency and carrying precision through environment-task-equipment multidimensional perception, combined with scene modeling and accurate positioning.

Description

Technical Field

[0001] This invention relates to the field of intelligent logistics automation technology, specifically to a method and system for controlling the handling of AGV stacker cranes for loading and unloading on production lines. Background Technology

[0002] The collaborative handling of AGVs and stacker cranes for loading and unloading on production lines is a core component of intelligent manufacturing logistics. However, traditional control methods have several technical limitations: path planning is based on preset maps and cannot dynamically avoid temporary obstacles such as tooling vehicles and personnel, easily leading to handling stagnation; the connection between AGV and stacker crane actions relies on fixed timing, lacking a real-time coordination mechanism, resulting in long loading and unloading waiting times; the end-point gripping uses rigid positioning, which is difficult to adapt to material posture deviations, such as offsets of ±10mm or more, easily leading to gripping failures or material damage; task scheduling does not consider the dynamic needs of the production line, and the priority of urgent materials cannot be flexibly adjusted, resulting in low overall flow efficiency. These problems restrict the flexibility and intelligence level of the production line logistics system, making it difficult to meet the needs of modern manufacturing with multiple varieties and small batches. Therefore, developing a handling control method for AGVs and stacker cranes for loading and unloading on production lines that can overcome the above defects is of significant practical importance. Summary of the Invention

[0003] To address the aforementioned technical shortcomings, this invention provides a method and system for controlling the handling of AGV stacker cranes on production lines, thereby improving the intelligence and efficiency of material handling on production lines.

[0004] This invention is achieved through the following technical solution: A method for controlling the handling of AGV stacker cranes during production line loading and unloading is provided, the method comprising the following steps: Step S10: Construct a multi-dimensional perception unit for environment, task, and device, and model and accurately locate the scene through laser SLAM, visual recognition, and RFID positioning; Step S20: Based on the improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices. Step S30: Use a fuzzy PID algorithm to adjust the AGV's travel speed and the stacker crane's lifting speed, and use a time axis synchronization mechanism to achieve seamless coordination between the two actions; Step S40: Construct an AGV stacker crane handling task priority scheduling model, optimize the handling sequence through digital twin pre-simulation, compare the pre-simulation and actual handling data, calculate the path deviation and energy consumption difference, and iteratively optimize the scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

[0005] Preferably, the step of constructing the environment-task-device multi-dimensional perception unit in step S10 includes: AGV Body Perception: A 2D laser SLAM radar is installed on the top of the AGV body, with a scanning angle of 360° and a ranging range of 0.1-30m, for environmental obstacle detection. An industrial vision camera is integrated at the front end of the AGV body, equipped with a 16mm fixed-focus lens and a ring LED light source, for identifying material QR codes and posture. Stacker crane positioning: An RFID tag with a UHF band is installed every 500mm along the vertical direction of the stacker crane's main column, i.e., along the movement trajectory of the stacker crane's lifting platform. The tag pre-stores the absolute height information corresponding to that position. The stacker crane's lifting platform, i.e., the movable platform that carries materials or forks, moves up and down along the guide rails of the column. An RFID reader is installed on its side. When the lifting platform moves up and down, the installed RFID reader can read the RFID tag information on the column in real time. Combined with the encoder data of the lifting platform itself, the information is calibrated to accurately obtain the current height position of the lifting platform with an error of ≤±3mm, providing a position reference for the coordinated action of the AGV and the stacker crane. End-effector sensing: The robotic gripper integrates a six-dimensional force sensor to detect contact forces during the gripping process and determine gripping stability. This includes X / Y / Z axis forces and torques. The X / Y / Z axis forces are linear forces acting on the robotic gripper in three orthogonal directions in a Cartesian coordinate system when gripping materials. The X-axis force is the force along the horizontal left-right direction of the robotic gripper, the Y-axis force is the force along the horizontal front-back direction of the robotic gripper, and the Z-axis force is the force along the vertical direction of the robotic gripper. These forces are mainly the gravitational reaction force of the material and the gripping clamping force. The values ​​of these forces directly reflect the tightness of the grip. The X / Y / Z axis torques are the torsional torques of the robotic gripper rotating around the three coordinate axes. The X-axis torque is the flipping torque around the horizontal left-right axis, the Y-axis torque is the flipping torque around the horizontal front-back axis, and the Z-axis torque is the rotational torque around the vertical axis. Data Synchronization Acquisition: All devices are connected via industrial Ethernet, using the Profinet protocol and precise time synchronization via IEEE1588 to ensure data timestamp error ≤5ms. LiDAR data, visual images, RFID positioning and force sensor data are transmitted to the edge controller in real time. Anti-interference design: The industrial vision camera is equipped with a dust cover and heating element to adapt to the dust and temperature difference environment of the workshop. The RFID tag is made of heat-resistant and oil-resistant material. The force sensor signal is filtered to remove mechanical vibration interference signal and can be filtered with 10Hz low-pass filter.

[0006] Preferably, the step S10, which involves scene modeling and precise positioning using laser SLAM, visual recognition, and RFID positioning, includes: Environmental map construction and updating: Based on laser SLAM technology, the backend is optimized using the GTSAM map optimization library to construct a 3D environmental map containing shelves, workstations and aisles. The map resolution is 5mm and the update frequency is 1Hz. Temporary obstacles are detected by the difference between point clouds in consecutive frames, and their positions and motion states (i.e., stationary / moving) are marked to update the obstacle information on the map. AGV positioning: The ICP iterative nearest point algorithm is used to match the laser point cloud with the map, and combined with the wheel odometer data fusion to achieve real-time positioning of the AGV. The positioning error is ≤±5mm and the update cycle is 50ms. When the laser signal is blocked, it automatically switches to visual positioning, that is, positioning based on QR code landmarks, to ensure positioning continuity. Material and equipment status identification: An industrial vision camera captures QR codes on the material surface, and the YOLOv8 model detects the QR code position, parsing the material specifications, weight, priority, and target workstation information. The material specifications are length × width × height, and the priority is set to 1-5 levels, with higher priority numbers being larger. The identification time is ≤20ms, and the accuracy is ≥99%. An RFID reader reads the height tag of the stacker crane lifting platform, combines it with encoder data calibration, obtains the real-time height, with an error of ≤±3mm, and synchronizes it to the stacker crane control system. A six-dimensional force sensor monitors the gripping contact force in real time, calculates the average and fluctuation values ​​of the gripping force, and determines whether the material is gripped securely. A fluctuation value ≤10N is considered stable.

[0007] Preferably, in step S20, based on improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and a real-time production line congestion index, and generates the optimal path based on these factors. The steps include: Path planning model construction: Improved A The algorithm framework uses the Manhattan distance from the starting point to the target point as the basic heuristic function, and introduces an energy consumption factor with a weight of 0.3 and a time factor with a weight of 0.7 to comprehensively evaluate the path cost. The energy consumption factor is calculated based on the AGV travel distance, the stacker crane lifting height, and the number of turns, while the time factor is calculated based on the AGV travel path length, the AGV preset speed, and the area congestion index. The AGV travel distance, stacker crane lifting height, number of turns, AGV travel path length, AGV preset speed, and area congestion index are transformed into calculable cost values. The AGV travel distance, stacker crane lifting height, and number of turns represent the energy consumption of the path, while the AGV travel path length, AGV preset speed, and area congestion index represent the efficiency of the path. Dynamic obstacle handling: Add an obstacle penalty term to the total cost calculation, using the formula P=W 障 Calculated as ×[1-(d / R)], where d is the actual distance from the previous path node to the obstacle center, R is the obstacle's penalty radius, which is equal to its width plus the set safety distance, and W...障 The dynamic weight of the obstacle is P, which is the obstacle penalty term. The obstacle weight is classified into levels: mobile device weight is 1.5, static temporary obstacle weight is 1.0, and fixed facility weight is 0.8. When the mobile device speed is ≥0.5m / s, the weight increases to 2.0, and high-speed moving objects are avoided first. The total cost = energy consumption + efficiency + the sum of obstacle penalty terms is used to quantify and score all possible paths. The path with the lowest total cost is the optimal path. Multi-segment path collaborative planning: For the AGV ground path, a smooth trajectory with continuous curvature is generated to reduce material swaying caused by sharp turns. The stacker crane lifting path adopts an S-shaped acceleration and deceleration curve to reduce mechanical impact. The end effector path plans the rotation angle and fine-tuning distance according to the material posture deviation to ensure accurate gripping. Set up a congestion avoidance mechanism: calculate the congestion index of each area in real time. The congestion index is set from 0 to 10, based on the number of devices and dwell time in the area in the past 30 seconds. When the congestion index is ≥7, route detour is triggered, and a suboptimal route is replanned to avoid regional congestion.

[0008] Preferably, step S30, which uses a fuzzy PID algorithm to adjust the AGV's travel speed and the stacker crane's lifting speed, and achieves the connection between their actions through a time axis synchronization mechanism, includes: Speed ​​Coordinated Control: Using a fuzzy PID algorithm, the inputs are the AGV position deviation and speed deviation. The AGV position deviation is the distance between the current position and the target path. The output is the PWM duty cycle of the drive wheel. The AGV travel speed is dynamically adjusted. The stacker crane lifting control is based on the difference between the target height and the current height. The speed of the lifting motor is adjusted by fuzzy PID. The time axis is synchronized by setting the time deviation between the AGV arriving at the picking point and the stacker crane descending to the position to ≤50ms. The action connection is achieved by pre-acceleration / deceleration compensation. End-effector adaptive control includes visual guidance positioning, force feedback adjustment, and anomaly handling mechanisms. In visual guidance positioning, the deviation between the center of the material's QR code and the center of the mechanical gripper is identified. When the deviation is > ±5mm or > ±1°, the end effector is controlled to make an initial adjustment. For force feedback adjustment, during the gripping contact stage, the PD control algorithm is used to perform attitude compensation based on the gripping force collected by the six-dimensional force sensor data. This includes X / Y axis translation, Z-axis height, rotation angle, and gripping stability judgment. The anomaly handling mechanism includes collision warning, gripping failure, and communication interruption.

[0009] Preferably, the step S40, which involves constructing an AGV stacker crane handling task priority scheduling model and optimizing the handling sequence through digital twin simulation, includes: Digital twin model construction: Based on Unity3D, a virtual scene of the production line is built, including digital twins of AGVs, stacker cranes, shelves and materials, etc. The physical parameters are consistent with the physical entities, and a control logic mapping is established so that the virtual equipment actions are synchronized with the control commands of the real equipment. Task timing simulation: Receive the set handling task, including the bill of materials, priority and time window, and simulate the task execution process in a virtual environment. Based on the simulation results, adjust the AGV departure time and stacker crane scheduling order to optimize the timing and reduce equipment waiting time. Data closed-loop optimization: Real-time collection of AGV stacker crane handling operation data, including path trajectory, speed curve, energy consumption and completion time, comparison of virtual and real data to calculate deviation, and use PPO reinforcement learning algorithm to iteratively optimize path planning weights and PID parameters with the goal of minimizing total handling time and energy consumption; Intelligent scheduling optimization: Task priorities are dynamically adjusted. When a task with a priority of level 5 is inserted, the current task sequence is re-planned to ensure that the response time for emergency tasks is ≤30s. Load balancing is achieved by allocating tasks according to the current load rate of each AGV stacker to avoid equipment overload, and the load rate is controlled at ≤80%.

[0010] Furthermore, to achieve the above objectives, the present invention also proposes a production line loading and unloading AGV stacker crane handling control system, which includes: Multi-source fusion perception and dynamic scene modeling and localization module: used to build a multi-dimensional perception unit for environment, task and equipment, and to model and accurately locate the scene through laser SLAM, visual recognition and RFID positioning; AGV stacker crane handling path planning module: used for improving A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices. AGV-Stacker Collaborative Control Module: Used to adjust the AGV travel speed and stacker lifting speed using a fuzzy PID algorithm, and to achieve seamless coordination between the two actions through a time axis synchronization mechanism; AGV stacker crane handling task priority scheduling and digital twin closed-loop optimization module: used to build AGV stacker crane handling task priority scheduling model, optimize handling sequence through digital twin pre-simulation, compare pre-simulation and actual handling data, calculate path deviation and energy consumption difference, and iteratively optimize scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

[0011] Furthermore, to achieve the above objectives, the present invention also proposes a production line loading and unloading AGV stacker crane handling control device, the device comprising: a memory, a processor, and programs such as a production line loading and unloading AGV stacker crane handling control algorithm stored in the memory and executable on the processor, wherein the production line loading and unloading AGV stacker crane handling control algorithm and other programs are the steps for implementing the production line loading and unloading AGV stacker crane handling control method described above.

[0012] In addition, to achieve the above objectives, the present invention also provides a computer program product, which includes programs such as a production line loading and unloading AGV stacker crane handling control algorithm. When the production line loading and unloading AGV stacker crane handling control algorithm and other programs are executed by a processor, they implement the production line loading and unloading AGV stacker crane handling control method described above.

[0013] The advantages and effects of this invention are: This invention proposes a handling control method and system for AGV stacker cranes used in production lines. It achieves temporary obstacle detection and material posture recognition through multi-source sensor fusion, solving the problem of rigid paths on traditional fixed maps. Simultaneously, it optimizes handling efficiency through intelligent planning, improving the AGV's handling capabilities. The algorithm combines energy consumption and time factors, and the dynamic path planning shortens the time for a single task and improves the throughput of congested areas. In addition, the fuzzy PID algorithm is used to adjust the AGV driving speed and the stacker crane lifting speed. The time axis synchronization mechanism realizes the connection between the two actions, which reduces the time synchronization error between the AGV and the stacker crane. Continuous optimization is achieved through digital twins, which improves the overall material flow efficiency and energy consumption efficiency of the production line. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of a material handling control method for an AGV stacker crane used for loading and unloading on a production line, according to the present invention.

[0016] Figure 2 This is a schematic diagram of the structure of an AGV stacker crane handling control system for loading and unloading materials on a production line according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1 As shown, in one embodiment of the present invention, a method for controlling the handling of AGV stacker cranes for loading and unloading on a production line includes the following steps: Step S10: Construct a multi-dimensional perception unit for environment, task, and device, and model and accurately locate the scene through laser SLAM, visual recognition, and RFID positioning.

[0019] Specifically, the steps in step S10 for constructing the environment-task-device multi-dimensional perception unit include: AGV Body Perception: A 2D laser SLAM radar is installed on the top of the AGV body, with a scanning angle of 360° and a ranging range of 0.1-30m, for environmental obstacle detection. An industrial vision camera is integrated at the front end of the AGV body, equipped with a 16mm fixed-focus lens and a ring LED light source, for identifying material QR codes and posture. Stacker crane positioning: An RFID tag in the UHF band is installed every 500mm along the vertical direction of the stacker crane's main column, i.e., along the movement trajectory of the stacker crane's lifting platform. The tag pre-stores the absolute height information corresponding to that position, such as 0.5m, 1.0m, or 1.5m from the stacker crane base. The main column serves as the vertical support frame of the stacker crane, typically consisting of 1-2 single or double-column metal structural components, such as square steel or channel steel, vertically fixed to the stacker crane's base or rails. Its height matches the maximum storage height of the production line's shelving. The stacker crane's lifting platform, a movable platform carrying materials or forks, moves up and down along the guide rails of the main column. RFID readers are installed on its sides. When the lifting platform rises or falls, the installed RFID readers can read the RFID tag information on the passing columns in real time. Combined with the lifting platform's own encoder data, calibration is performed to accurately obtain the current height position of the lifting platform, with an error ≤ ±3mm. This provides a positional reference for the coordinated action of the AGV and the stacker crane, such as ensuring the stacker crane accurately descends to the material-picking height when the AGV arrives. End-effector sensing: The robotic gripper integrates a six-dimensional force sensor to detect contact forces during the gripping process and determine gripping stability. This includes X / Y / Z-axis forces and torques. The X / Y / Z-axis forces are linear forces acting on the robotic gripper in three orthogonal directions within a Cartesian coordinate system when gripping material. The X-axis force is the force along the horizontal left-right direction of the gripper, such as the lateral thrust caused by material misalignment. The Y-axis force is the force along the horizontal front-back direction of the gripper, such as the frictional force between the material and the gripper during gripping. The Z-axis force is the force perpendicular to the gripper, mainly consisting of the material's gravitational reaction force and the gripping clamping force. The values ​​of these forces directly reflect the gripping stability. The tightness of the gripper is important. For example, if the Z-axis force is too small, the material may slip off, while if it is too large, it may damage the material. The X / Y / Z-axis torques are the torsional torques of the mechanical gripper rotating around the three coordinate axes. The X-axis torque is the flipping torque around the horizontal left and right axes, such as the tendency of the material to tilt down / up due to the center of gravity being off-center. The Y-axis torque is the flipping torque around the horizontal front and back axes, such as the tendency of the material to tilt forward / backward due to the front end being heavier. The Z-axis torque is the rotational torque around the vertical axis, such as the torsional tendency caused by the material not being concentric with the mechanical gripper. The magnitude of the torque reflects the stability of the material's posture. When the torque on a certain axis is too large, it indicates that the material is tilted or eccentric, which will lead to an imbalance in the gripping. Data Synchronization Acquisition: All devices are connected via industrial Ethernet, using the Profinet protocol and precise time synchronization via IEEE1588 to ensure data timestamp error ≤5ms. LiDAR data, visual images, RFID positioning and force sensor data are transmitted to the edge controller in real time. Anti-interference design: The industrial vision camera is equipped with a dust cover and heating element to adapt to the dust and temperature difference environment of the workshop. The RFID tag is made of heat-resistant and oil-resistant material, such as being able to withstand temperatures of -40℃ to 85℃, to ensure positioning stability. The force sensor signal is filtered to remove mechanical vibration interference signals, and a 10Hz low-pass filter can be used.

[0020] Specifically, step S10, which involves scene modeling and precise positioning using laser SLAM, visual recognition, and RFID positioning, includes: Environmental map construction and updating: Based on laser SLAM technology, the backend is optimized using the GTSAM map optimization library to construct a 3D environmental map containing shelves, workstations, and aisles. The map resolution is 5mm and the update frequency is 1Hz. Temporary obstacles, such as suddenly appearing work vehicles, are detected by the difference between point clouds in previous and next frames. Their positions and motion states, i.e., stationary / moving, are marked, and the map obstacle information is updated. AGV positioning: The ICP iterative nearest point algorithm is used to match the laser point cloud with the map, and combined with the wheel odometer data fusion to achieve real-time positioning of the AGV. The positioning error is ≤±5mm and the update cycle is 50ms. When the laser signal is blocked, such as in densely packed shelving areas, it automatically switches to visual positioning, that is, positioning based on QR code landmarks, to ensure positioning continuity. Material and equipment status identification: An industrial vision camera captures QR codes on the material surface, and the YOLOv8 model detects the QR code position, parsing the material specifications, weight, priority, and target workstation information. The material specifications are length × width × height, and the priority is set to 1-5 levels, with higher priority numbers being larger. The identification time is ≤20ms, and the accuracy is ≥99%. An RFID reader reads the height tag of the stacker crane lifting platform, combines it with encoder data calibration, obtains the real-time height, with an error of ≤±3mm, and synchronizes it to the stacker crane control system. A six-dimensional force sensor monitors the gripping contact force in real time, calculates the average and fluctuation values ​​of the gripping force, and determines whether the material is gripped securely. A fluctuation value ≤10N is considered stable.

[0021] Step S20: Based on the improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices.

[0022] Specifically, in step S20, based on the improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and a real-time production line congestion index, and generates the optimal path based on these factors. The steps include: Path planning model construction: Improved A The algorithm framework uses the Manhattan distance from the starting point to the target point as the basic heuristic function, and introduces an energy consumption factor with a weight of 0.3 and a time factor with a weight of 0.7 to comprehensively evaluate the path cost. The energy consumption factor is calculated based on the AGV travel distance, the stacker crane lifting height, and the number of turns, while the time factor is calculated based on the AGV travel path length, the AGV preset speed, and the area congestion index. The AGV travel distance, stacker crane lifting height, number of turns, AGV travel path length, AGV preset speed, and area congestion index are transformed into calculable cost values. The AGV travel distance, stacker crane lifting height, and number of turns represent the energy consumption of the path, while the AGV travel path length, AGV preset speed, and area congestion index represent the efficiency of the path. Dynamic obstacle handling: Add an obstacle penalty term to the total cost calculation, using the formula P=W 障 Calculated as ×[1-(d / R)], where d is the actual distance from the previous path node to the obstacle center, R is the obstacle's penalty radius, which is equal to its width plus the set safety distance, and W... 障The dynamic weight of obstacles is P, which is the obstacle penalty term. Obstacle weights are classified into levels: mobile devices have a weight of 1.5, such as other AGVs; static temporary obstacles have a weight of 1.0, such as tool carts; and fixed facilities have a weight of 0.8, such as shelves. When the speed of mobile devices is ≥0.5m / s, the weight increases to 2.0, prioritizing the avoidance of high-speed moving objects. The total cost is calculated using the formula: total cost = energy consumption + efficiency + the sum of obstacle penalties. All possible paths are quantitatively scored, and the path with the lowest total cost is the optimal path. Multi-segment path collaborative planning: For the AGV ground path, a smooth trajectory with continuous curvature is generated to reduce material swaying caused by sharp turns, with a maximum curvature of 0.5 rad / m. The stacker crane lifting path adopts an S-shaped acceleration and deceleration curve, with a lifting speed of 0-0.8 m / s and an acceleration of ≤0.5 m / s², reducing mechanical impact. The end effector path is planned according to the material posture deviation, with a rotation angle accuracy of ±0.5° and a fine-tuning distance of ±5 mm to ensure accurate gripping. Set up a congestion avoidance mechanism: Calculate the congestion index of each area in real time. The congestion index is set from 0 to 10. It is calculated based on the number of devices and dwell time in the area in the past 30 seconds. When the congestion index is ≥7, route detour is triggered and a suboptimal route is replanned. The cost increases by ≤10% to avoid regional congestion.

[0023] Step S30: Use a fuzzy PID algorithm to adjust the AGV travel speed and the stacker crane lifting speed, and use a time axis synchronization mechanism to achieve the connection between the two actions.

[0024] Specifically, step S30, which uses a fuzzy PID algorithm to adjust the AGV's travel speed and the stacker crane's lifting speed, and achieves the connection between their actions through a time axis synchronization mechanism, includes the following steps: Speed ​​Coordinated Control: Using a fuzzy PID algorithm, the inputs are the AGV position deviation and speed deviation. The AGV position deviation is the distance between the current position and the target path. The output is the PWM duty cycle of the drive wheel, which dynamically adjusts the AGV's travel speed. The AGV travel speed adjustment range is 0-1.5m / s. The stacker crane lifting control is based on the difference between the target height and the current height. The lifting motor speed is adjusted by fuzzy PID, with an adjustment range of 0-0.8m / s and a positioning accuracy of ±3mm. Time axis synchronization is achieved by setting the time deviation between the AGV arriving at the picking point and the stacker crane descending to the bottom to ≤50ms. Action connection is achieved through pre-acceleration / deceleration compensation. End-effector adaptive control includes visual guidance positioning, force feedback adjustment, and anomaly handling mechanisms. In visual guidance positioning, the deviation between the center of the material's QR code and the center of the mechanical gripper is identified. When the deviation is >±5mm or >±1°, the end effector is controlled to make an initial adjustment. For force feedback adjustment, during the gripping contact phase, the PD control algorithm is used for attitude compensation based on the gripping force acquired from the six-dimensional force sensor data. This includes X / Y axis translation, Z-axis height, rotation angle, and gripping stability assessment. During X / Y axis translation, the gain K is adjusted. p =2.5, K d =0.8, fine-tuning range ±5mm, Z-axis height adjustment gain K p =3.0, K d =1.0, ensuring consistent gripping depth, adjust gain K during rotation angle. p =1.5, K d =0.5, fine-tuning range ±2°. In the grasping stability judgment, when the six-dimensional force sensor detects that the grasping force is stable, if the fluctuation is ≤5N within 300ms, the grasping is confirmed to be successful; otherwise, a retry mechanism is triggered, and it can be triggered a maximum of 3 times. The abnormal handling mechanism includes collision warning, grasping failure, and communication interruption. The collision warning is triggered when the lidar detects an obstacle within 0.5m ahead, and deceleration is immediately triggered, reducing the speed to 0.2m / s within 100ms. If it continues to approach, a stop alarm is triggered. Grasping failure is triggered when there are 3 consecutive grasping failures, and the system is automatically reported, and a material position deviation image is sent to the monitoring terminal. Communication interruption is triggered by using a local cache path to ensure that the AGV and the stacker crane safely stop after completing the current action and wait for communication to be restored.

[0025] Step S40: Construct an AGV stacker crane handling task priority scheduling model, optimize the handling sequence through digital twin pre-simulation, compare the pre-simulation and actual handling data, calculate the path deviation and energy consumption difference, and iteratively optimize the scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

[0026] Specifically, step S40, which involves constructing an AGV stacker crane handling task priority scheduling model and optimizing the handling sequence through digital twin simulation, includes the following steps: Digital twin model construction: Based on Unity3D, a virtual scene of the production line is built, including digital twins of AGVs, stacker cranes, shelves and materials, etc. The physical parameters are consistent with the real ones, the size error is ≤1%, and a control logic mapping is established so that the virtual equipment actions are synchronized with the control commands of the real equipment, with a simulation accuracy of ≤5ms. Task timing simulation: Receive the set handling task, including the bill of materials, priority and time window, and simulate the task execution process in a virtual environment. Based on the simulation results, adjust the AGV departure time and stacker crane scheduling order to optimize the timing and reduce equipment waiting time. The optimized waiting time is ≤10s. Data closed-loop optimization: Real-time collection of AGV stacker crane handling operation data, including path trajectory, speed curve, energy consumption and completion time, comparison of virtual and real data to calculate deviations. Compared with real data, the path deviation is set to ≤3mm, energy consumption difference to ≤5%, and time deviation to ≤2%. The PPO reinforcement learning algorithm is adopted to iteratively optimize the path planning weight and PID parameters with the goal of minimizing total handling time and energy consumption, and iterates once every 1000 tasks. Intelligent scheduling optimization: Task priorities are dynamically adjusted. When a task with a priority of level 5 is inserted, the current task sequence is re-planned to ensure that the response time for emergency tasks is ≤30s. Load balancing is achieved by allocating tasks according to the current load rate of each AGV stacker to avoid equipment overload, and the load rate is controlled at ≤80%.

[0027] In addition, such as Figure 2 As shown, in one embodiment of the present invention, a production line loading and unloading AGV stacker crane handling control system is proposed. The system includes: Multi-source fusion perception and dynamic scene modeling and localization module: used to build a multi-dimensional perception unit for environment, task and equipment, and to model and accurately locate the scene through laser SLAM, visual recognition and RFID positioning; AGV stacker crane handling path planning module: used for improving A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices. AGV-Stacker Collaborative Control Module: Used to adjust the AGV travel speed and stacker lifting speed using a fuzzy PID algorithm, and to achieve seamless coordination between the two actions through a time axis synchronization mechanism; AGV stacker crane handling task priority scheduling and digital twin closed-loop optimization module: used to build AGV stacker crane handling task priority scheduling model, optimize handling sequence through digital twin pre-simulation, compare pre-simulation and actual handling data, calculate path deviation and energy consumption difference, and iteratively optimize scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

[0028] This application provides a production line loading / unloading AGV stacker crane handling control system, which employs a production line loading / unloading AGV stacker crane handling control method described in the above embodiments. This system solves the technical problems of rigid path planning, low coordination efficiency, and poor gripping adaptability in traditional AGV stacker crane handling. Compared with the prior art, the beneficial effects of the production line loading / unloading AGV stacker crane handling control system provided in this application are the same as those of the production line loading / unloading AGV stacker crane handling control method described in the above embodiments. Furthermore, other technical features of the production line loading / unloading AGV stacker crane handling control system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0029] This application provides a production line loading and unloading AGV stacker crane handling control device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the production line loading and unloading AGV stacker crane handling control method in the above embodiment 1.

[0030] In one embodiment of the present invention, a production line loading / unloading AGV stacker crane handling control device may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), etc., and fixed terminals such as digital TVs, desktop computers, etc. The described production line loading / unloading AGV stacker crane handling control device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments in this application.

[0031] The aforementioned production line loading / unloading AGV stacker crane handling control device may include a processing system (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage system into a machine-readable storage medium (RAM). The machine-readable storage medium also stores various programs and data required for the operation of the production line loading / unloading AGV stacker crane handling control device. The processing system, the read-only memory, and the machine-readable storage medium are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus. Typically, the following systems can be connected to the I / O interface: input systems including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output systems including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage systems including, for example, magnetic tapes, hard disks, etc.; and communication systems. The communication system allows the production line loading / unloading AGV stacker crane handling control device to communicate wirelessly or wiredly with other devices to exchange data. Although the diagram shows a production line AGV stacker crane handling control system with various systems, it should be understood that implementation or having all of the systems shown is not required. More or fewer systems may be implemented alternatively.

[0032] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication system, or installed from a storage system, or installed from a read-only memory. When the computer program is executed by a processing system, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0033] This application provides a production line loading / unloading AGV stacker crane handling control device, which employs a production line loading / unloading AGV stacker crane handling control method described in the above embodiments. This method solves the technical problems of rigid path planning, low coordination efficiency, and poor gripping adaptability in traditional AGV stacker crane handling. Compared with the prior art, the beneficial effects of the production line loading / unloading AGV stacker crane handling control device provided in this application are the same as those of the production line loading / unloading AGV stacker crane handling control method described in the above embodiments. Furthermore, other technical features of this production line loading / unloading AGV stacker crane handling control device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0034] The various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0035] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described production line loading and unloading AGV stacker crane handling control method.

[0036] The computer program product provided in this application can solve the technical problems of rigid path planning, low coordination efficiency, and poor gripping adaptability in traditional AGV stacker crane handling. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the AGV stacker crane handling control method for production line loading and unloading provided in the above embodiments, and will not be repeated here.

[0037] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for controlling the handling of AGV stacker cranes for loading and unloading on a production line, characterized in that, The method includes the following steps: Step S10: Construct a multi-dimensional perception unit for environment, task, and device, and model and accurately locate the scene through laser SLAM, visual recognition, and RFID positioning; Step S20: Based on the improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices. Step S30: Use a fuzzy PID algorithm to adjust the AGV's travel speed and the stacker crane's lifting speed, and use a time axis synchronization mechanism to achieve seamless coordination between the two actions; Step S40: Construct an AGV stacker crane handling task priority scheduling model, optimize the handling sequence through digital twin pre-simulation, compare the pre-simulation and actual handling data, calculate the path deviation and energy consumption difference, and iteratively optimize the scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

2. The method for controlling the handling of AGV stacker cranes for loading and unloading on a production line according to claim 1, characterized in that, The step of constructing the environment-task-device multi-dimensional perception unit in step S10 includes: AGV Body Perception: A 2D laser SLAM radar is installed on the top of the AGV body for environmental obstacle detection, and an industrial vision camera is integrated at the front end of the AGV body for recognizing material QR codes and posture. Stacker crane positioning: A UHF RFID tag is installed every 500mm along the vertical direction of the stacker crane's main column, i.e., along the movement trajectory of the stacker crane's lifting platform. The tag pre-stores the absolute height information corresponding to that position. The stacker crane's lifting platform moves up and down along the guide rail of the column. An RFID reader is installed on its side. When the lifting platform moves up and down, the installed RFID reader reads the RFID tag information on the column in real time. Combined with the encoder data of the lifting platform itself, the information is calibrated to accurately obtain the current height position of the lifting platform. End-effector sensing: The robotic gripper integrates a six-dimensional force sensor to detect contact forces during the gripping process and determine gripping stability. This includes X / Y / Z axis forces and torques. The X / Y / Z axis forces are linear forces acting on the robotic gripper in three orthogonal directions in a Cartesian coordinate system when gripping materials. The X-axis force is the force along the horizontal left-right direction of the robotic gripper, the Y-axis force is the force along the horizontal front-back direction of the robotic gripper, and the Z-axis force is the force along the vertical direction of the robotic gripper. The X / Y / Z axis torques are the torsional torques of the robotic gripper rotating around the three coordinate axes. The X-axis torque is the flipping torque around the horizontal left-right axis, the Y-axis torque is the flipping torque around the horizontal front-back axis, and the Z-axis torque is the rotational torque around the vertical axis. Data Synchronous Acquisition: All devices are connected via industrial Ethernet, using the Profinet protocol and precise time synchronization via IEEE1588. LiDAR data, visual images, RFID positioning, and force sensor data are transmitted to the edge controller in real time. Anti-interference design: The industrial vision camera is equipped with a dust cover and heating element to adapt to the dust and temperature difference environment of the workshop. The RFID tag is made of heat-resistant and oil-resistant material. The force sensor signal is filtered to remove mechanical vibration interference signals.

3. The method for controlling the handling of AGV stacker cranes for loading and unloading on a production line according to claim 1, characterized in that, The steps in step S10, which involve scene modeling and precise positioning using laser SLAM, visual recognition, and RFID positioning, include: Environmental map construction and updating: Based on laser SLAM technology, the backend is optimized using the GTSAM map optimization library to construct a 3D environmental map containing shelves, workstations, and aisles. Temporary obstacles are detected by the difference between point clouds in previous and next frames, and their positions and motion states (i.e., stationary / moving) are marked to update the map obstacle information. AGV positioning: The ICP iterative nearest point algorithm is used to match the laser point cloud with the map, and combined with wheel odometer data fusion to achieve real-time AGV positioning. When the laser signal is blocked, it automatically switches to visual positioning, that is, positioning based on QR code landmarks. Material and equipment status identification: An industrial vision camera captures the QR code on the material surface, and the YOLOv8 model detects the QR code position, parses the material specifications, weight, priority, and target workstation information. The material specifications are length × width × height, and the priority is set to 1-5 levels. The RFID reader reads the height tag of the stacker crane lifting platform, and combined with encoder data calibration, obtains the real-time height and synchronizes it to the stacker crane control system. The six-dimensional force sensor monitors the gripping contact force in real time, calculates the average and fluctuation values ​​of the gripping force, and determines whether the material is gripped securely.

4. The method for controlling the handling of AGV stacker cranes for loading and unloading on a production line according to claim 1, characterized in that, In step S20, based on improved A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and a real-time production line congestion index, and generates the optimal path based on these factors. The steps include: Path planning model construction: Improved A The algorithm framework uses the Manhattan distance from the starting point to the target point as the basic heuristic function, and introduces an energy consumption factor with a weight of 0.3 and a time factor with a weight of 0.7 to comprehensively evaluate the path cost. The energy consumption factor is calculated based on the AGV travel distance, the stacker crane lifting height, and the number of turns, while the time factor is calculated based on the AGV travel path length, the AGV preset speed, and the area congestion index. The AGV travel distance, stacker crane lifting height, number of turns, AGV travel path length, AGV preset speed, and area congestion index are transformed into calculable cost values. The AGV travel distance, stacker crane lifting height, and number of turns represent the energy consumption of the path, while the AGV travel path length, AGV preset speed, and area congestion index represent the efficiency of the path. Dynamic obstacle handling: Add an obstacle penalty term to the total cost calculation, using the formula P=W 障 Calculated as ×[1-(d / R)], where d is the actual distance from the previous path node to the obstacle center, R is the obstacle's penalty radius, which is equal to its width plus the set safety distance, and W... 障 The dynamic weight of the obstacle is P, which is the obstacle penalty term. The obstacle weight is classified into levels: mobile device weight is 1.5, static temporary obstacle weight is 1.0, and fixed facility weight is 0.

8. When the mobile device speed is ≥0.5m / s, the weight increases to 2.

0. All paths are quantitatively scored using the formula Total Cost = Energy Consumption + Efficiency + Sum of Obstacle Penalties. The path with the lowest total cost is the optimal path. Multi-segment path collaborative planning: For the AGV ground path, the material swaying is reduced by generating a smooth trajectory with continuous curvature. The stacker crane lifting path adopts an S-shaped acceleration and deceleration curve. The end effector path plans the rotation angle and fine-tuning distance according to the material posture deviation. Set up a congestion avoidance mechanism: calculate the congestion index of each area in real time. The congestion index is set from 0 to 10, based on the number of devices and dwell time in the area in the past 30 seconds. When the congestion index is ≥7, route detour is triggered and a suboptimal route is replanned.

5. The method for controlling the handling of AGV stacker cranes for loading and unloading on a production line according to claim 1, characterized in that, The step S30, which uses a fuzzy PID algorithm to adjust the AGV's travel speed and the stacker crane's lifting speed, and achieves the connection between their actions through a time axis synchronization mechanism, includes the following steps: Speed ​​Coordinated Control: Using a fuzzy PID algorithm, the inputs are the AGV position deviation and speed deviation. The AGV position deviation is the distance between the current position and the target path. The output is the PWM duty cycle of the drive wheel. The AGV travel speed is dynamically adjusted. The stacker crane lifting control is based on the difference between the target height and the current height. The speed of the lifting motor is adjusted by fuzzy PID. The time axis is synchronized by setting the time deviation between the AGV arriving at the picking point and the stacker crane descending to the position to ≤50ms. The action connection is achieved by pre-acceleration / deceleration compensation. End-effector adaptive control includes visual guidance positioning, force feedback adjustment, and anomaly handling mechanisms. In visual guidance positioning, the deviation between the center of the material's QR code and the center of the mechanical gripper is identified. When the deviation is > ±5mm or > ±1°, the end effector is controlled to make an initial adjustment. For force feedback adjustment, during the gripping contact stage, the PD control algorithm is used to perform attitude compensation based on the gripping force collected by the six-dimensional force sensor data. This includes X / Y axis translation, Z-axis height, rotation angle, and gripping stability judgment. The anomaly handling mechanism includes collision warning, gripping failure, and communication interruption.

6. The method for controlling the handling of AGV stacker cranes for loading and unloading on a production line according to claim 1, characterized in that, The steps in step S40, which involve constructing an AGV stacker crane handling task priority scheduling model and optimizing the handling sequence through digital twin simulation, include: Digital twin model construction: Based on Unity3D, a virtual scene of the production line is built, including digital twins of AGVs, stacker cranes, shelves and materials. The physical parameters are consistent with the physical entities, and a control logic mapping is established so that the virtual equipment actions are synchronized with the control commands of the real equipment. Task sequence rehearsal: Receive the set handling task, including the bill of materials, priority and time window, rehearse the task execution process in a virtual environment, and adjust the AGV departure time and stacker crane scheduling order based on the rehearsal results; Data closed-loop optimization: Real-time collection of AGV stacker crane handling operation data, including path trajectory, speed curve, energy consumption and completion time, comparison of virtual and real data to calculate deviation, and use PPO reinforcement learning algorithm to iteratively optimize path planning weights and PID parameters with the goal of minimizing total handling time and energy consumption; Intelligent scheduling optimization: Task priorities are dynamically adjusted. When a task with a priority of level 5 is inserted, the current task sequence is re-planned, load is balanced, and tasks are allocated according to the current load rate of each AGV stacker.

7. A material handling control system for an AGV stacker crane on a production line, characterized in that, include: Multi-source fusion perception and dynamic scene modeling and localization module: used to build a multi-dimensional perception unit for environment, task and equipment, and to model and accurately locate the scene through laser SLAM, visual recognition and RFID positioning; AGV stacker crane handling path planning module: used for improving A The algorithm dynamically plans collision-free paths, designs dynamic obstacle weight factors and real-time production line congestion indices, and generates the optimal path based on the dynamic obstacle weight factors and real-time production line congestion indices. AGV stacker collaborative control module: used to adjust the AGV travel speed and stacker lifting speed using a fuzzy PID algorithm, and to achieve the connection between the two actions through a time axis synchronization mechanism; AGV stacker crane handling task priority scheduling and digital twin closed-loop optimization module: used to build AGV stacker crane handling task priority scheduling model, optimize handling sequence through digital twin pre-simulation, compare pre-simulation and actual handling data, calculate path deviation and energy consumption difference, and iteratively optimize scheduling strategy and AGV stacker crane control parameters through reinforcement learning.

8. A material handling control device for an AGV stacker crane on a production line, characterized in that, include: The system includes a memory, a processor, and a production line loading / unloading AGV stacker crane handling control program stored in the memory and executable on the processor. When the production line loading / unloading AGV stacker crane handling control program is executed by the processor, it implements a production line loading / unloading AGV stacker crane handling control method as described in any one of claims 1 to 6.

9. A computer program product, characterized in that, The invention includes a production line loading and unloading AGV stacker crane handling control program, which, when executed by a processor, implements a production line loading and unloading AGV stacker crane handling control method as described in any one of claims 1 to 6.

Images