Automatic goods warehousing method and system for intelligent warehousing

Abstract

The invention discloses an automatic goods warehousing method and system for intelligent warehousing, and belongs to the technical field of intelligent warehousing. According to the method, dynamic event information such as cargo elevator fault signals and robot position conflict detection data is obtained through a real-time data interface, a dynamic path self-adaptive scheduling mechanism is constructed based on a path planning algorithm, and a standby path set is preset; when a dynamic event is detected, the edge calculation unit is triggered to complete a standby path calculation strategy within a system preset time threshold value, and finally the storage robot is controlled to execute storage work according to a planned path. The system comprises a data sensing module, a path planning module, an edge calculation unit and an execution control module, and the modules cooperate to realize real-time path optimization under a dynamic event. According to the method, the problems of insufficient dynamic scene adaptability and low equipment cooperation efficiency in traditional storage are effectively solved, the storage efficiency is improved, and the energy consumption is reduced.

Description

Technical Field

[0001] The present invention relates to an automatic goods warehousing method and system for intelligent warehousing, and belongs to the technical field of intelligent warehousing. Background Art

[0002] With the vigorous development of e-commerce logistics and intelligent manufacturing, intelligent warehousing systems have become the core link in improving supply chain efficiency. At present, the automated warehousing of goods, as a key process in intelligent warehousing, faces technical challenges such as insufficient adaptability to dynamic scenarios and low equipment coordination efficiency. Specifically, in traditional warehousing, warehouse robots need to rely on manual scheduling of freight elevators to transport goods across floors, resulting in independent operation of freight elevators and robots and data incommunicability; static path planning is difficult to cope with dynamic events, and operation continuity is poor; in the warehousing process, data of robots, freight elevators, conveyor lines and other equipment are scattered in different systems, lacking a unified collaborative scheduling model. In order to solve the above problems, the present invention proposes an automated warehousing method and system for intelligent warehousing. Summary of the Invention

[0003] To address the problems of insufficient adaptability to dynamic scenarios, low equipment coordination efficiency, and frequent manual intervention in existing intelligent warehousing warehousing processes, the present invention provides an automated cargo warehousing method and system for intelligent warehousing. Through multi-source data fusion, dynamic path planning, and equipment coordinated scheduling, this method achieves real-time linkage between freight elevators and robots, improving warehousing efficiency and reducing energy consumption. This invention is implemented using the following technical solutions.

[0004] On the one hand, the present invention provides a method for automated warehousing of goods for intelligent warehousing, wherein the automated warehousing method is applied to an automated warehousing system, which is used to control the warehousing of a warehousing robot and includes multiple edge computing units, and is characterized in that the automated warehousing method includes: obtaining dynamic event information in the goods warehousing scene through a real-time data interface; when a dynamic event is detected, selecting at least one edge computing unit to jointly generate at least one backup path for the warehousing robot based on the dynamic event information and the position information of the warehousing robot and the surrounding image information collected by the warehousing robot, wherein the duration of the joint processing of the selected edge computing units is shorter than the system preset duration; controlling the warehousing robot to perform the warehousing operation according to one of the backup paths.

[0005] Furthermore, the real-time data interface includes an industrial communication protocol interface for data interaction with the freight elevator controller and the robot controller, and the dynamic event information includes freight elevator fault signals and robot position conflict detection data.

[0006] Furthermore, the freight elevator fault signal further includes: obtaining the operating status data of the freight elevator PLC through the industrial Ethernet interface, and the operating status data includes the freight elevator position, load weight, and door switch status.

[0007] Furthermore, the robot position conflict detection data further includes: obtaining the three-dimensional coordinate data of each warehouse robot in real time through the positioning system, and the positioning accuracy of the positioning system meets the positioning error requirements preset by the system; obtaining the robot position conflict data through the deployed lidar sensor, and the scanning frequency and angular resolution of the lidar meet the detection accuracy requirements preset by the system.

[0008] Furthermore, after obtaining the robot position conflict data through the positioning system and the lidar sensor, the method further includes: performing timestamp synchronization on the obtained multi-source heterogeneous data, wherein the multi-source heterogeneous data includes the data obtained by the positioning system and the data obtained by the lidar, and the error of the timestamp synchronization meets the synchronization accuracy requirements preset by the system; using a data fusion algorithm to process the robot position data to eliminate the random error of the positioning system; and calculating the conflict characteristic parameters such as the relative speed and minimum approach distance of adjacent robots based on the processed position data.

[0009] Furthermore, the dynamic path adaptive scheduling mechanism based on the path planning algorithm further includes: constructing a basic path planning model based on a graph search algorithm; introducing time window constraints to construct a multi-robot path conflict avoidance model; and optimizing and training the path planning model based on a machine learning algorithm.

[0010] Furthermore, the preset backup path set further includes: pre-generating at least three backup paths with different priorities based on the three-dimensional map of the warehouse; the path priority is comprehensively determined by at least two of the path length, energy consumption coefficient, equipment usage frequency, and congestion probability; when the main path is occupied or fails, the backup path is selected in order of priority.

[0011] Furthermore, the edge computing unit executes a backup path calculation strategy, which further includes: parallel calculation of multiple candidate paths based on the multi-core processor of the edge computing unit; using a heuristic search algorithm to complete the optimal path screening within a time threshold preset by the system; when computing resources are insufficient, offloading part of the computing tasks to the cloud server.

[0012] Furthermore, the control of the warehouse robot to perform warehousing operations according to the planned path further includes: generating robot motion control instructions based on the path planning results, the instructions including speed curve, steering angle, and acceleration parameters; transmitting the control instructions to the warehouse robot in real time through the industrial wireless communication network; and using a closed-loop control algorithm to control the robot's motion trajectory, and the trajectory tracking error meets the system's preset accuracy requirements.

[0013] On the other hand, the present invention provides an automated cargo warehousing system for intelligent warehousing, comprising: a data perception module, for obtaining freight elevator fault signals and robot position conflict data through sensors; a path planning module, with a built-in dynamic path adaptive scheduling mechanism and a backup path set, wherein the backup path set includes at least one backup path; an edge computing unit, for completing the backup path calculation strategy within a preset time threshold range when a dynamic event is detected; an execution control module, for controlling the warehousing robot to perform warehousing operations according to the planned path; wherein the data perception module, path planning module, edge computing unit and execution control module work together to achieve real-time path optimization under dynamic events.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] (1) Continuously obtain dynamic event information in the cargo entry scenario through a real-time data interface. When a dynamic event is detected, the edge computing unit collaboration mechanism is immediately activated, avoiding the transmission delay of the traditional central processing mode;

[0016] (2) Based on the dynamic event type and real-time environmental data, the edge computing unit collaboratively generates at least one backup path, effectively reducing the impact of dynamic events on warehousing operations and ensuring the continuity of the goods warehousing process;

[0017] (3) The generated alternative path comprehensively considers factors such as path length and energy consumption coefficient, so that the warehouse robot consumes less energy when performing warehousing operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Shown is a workflow diagram of a method for automated goods entry for intelligent warehousing;

[0019] Figure 2 The figure shows a schematic diagram of the planned path;

[0020] Figure 3 The figure shows a structural diagram of an automated goods storage system for intelligent warehousing. DETAILED DESCRIPTION

[0021] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0022] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0024] This embodiment provides a method for automatically storing goods for intelligent warehousing. Figure 1 As shown, the automated warehousing method is applied to an automated warehousing system, which is used to control warehousing robots and includes multiple edge computing units. The automated warehousing method includes:

[0025] S101. Acquire dynamic event information in the goods warehousing scenario through a real-time data interface;

[0026] S102. When a dynamic event is detected, select at least one edge computing unit to jointly generate at least one backup path for the warehouse robot based on the dynamic event information, the location information of the warehouse robot, and surrounding image information collected by the warehouse robot, wherein the duration of the joint processing by the selected edge computing units is shorter than a system preset duration;

[0027] S103: Control the storage robot to perform warehousing operations according to one of the backup paths.

[0028] This embodiment obtains dynamic event information such as freight elevator fault signals and robot position conflict detection data through a real-time data interface, builds a dynamic path adaptive scheduling mechanism based on the path planning algorithm, and presets a set of backup paths. When a dynamic event is detected, the edge computing unit is triggered to complete the backup path calculation strategy within the system's preset time threshold, and ultimately controls the warehouse robot to perform warehousing operations according to the planned path.

[0029] In this embodiment, dynamic event information is acquired through a real-time data interface, and the operating status of the freight elevator and robots is collected in real time using a monitoring system. Specifically, sensors are deployed at key locations on the freight elevator to monitor for abnormal conditions during operation. If fault signals such as vibration amplitude exceeding a threshold, abnormal current increase, or unusual motor noise are detected, the sensors immediately generate an alarm signal. Simultaneously, positioning devices and lidar within the warehouse continuously monitor the robots' spatial positions. If two robots are about to occupy the same path at the same time, or if the distance between them falls below a safety threshold, the system generates conflict warning data in real time and transmits it to the central control system. Furthermore, the system pre-plans backup paths for the robots, establishing a dynamic path redundancy mechanism. Engineers use an intelligent path planning algorithm based on the three-dimensional warehouse layout to pre-generate at least one backup path. Backup path planning comprehensively considers multiple factors, including path length, energy consumption, equipment usage frequency, and congestion probability, and prioritizes each path based on these factors. For example, while the primary path is the shortest, backup path A can avoid high-congestion risk sections, while backup path B is suitable for scenarios where the elevator fails.

[0030] When a freight elevator malfunction or a robot collision risk is detected, the system immediately triggers the edge computing unit, serving as the local intelligent control core. Based on an embedded computing architecture, this unit initiates a backup path calculation strategy within a preset time threshold. Based on the urgency of the dynamic event and the actual on-site situation, it quickly selects the optimal path from a set of backup paths. For example, if a freight elevator malfunctions, the system prioritizes routes that do not rely on the elevator. If a robot path conflict is detected, it selects a path that allows for rapid two-way obstacle avoidance. This local computing mode avoids data transmission latency from the cloud, achieving millisecond-level response through real-time decision-making at the edge node. After path planning is complete, the system sends a control data packet containing trajectory coordinates, speed parameters, and steering instructions to the robot, which then automatically reconstructs the trajectory. Simultaneously, the system monitors the robot's position in real time using a fusion of laser SLAM and inertial navigation technology. Using a PID closed-loop control algorithm, the system dynamically adjusts motor output parameters to keep trajectory tracking error within millimeter limits, ensuring the accuracy and safety of warehousing operations.

[0031] The present invention integrates real-time data acquisition, dynamic path planning, edge computing, and equipment collaborative control technology systems: through industrial sensors and communication protocols, a global monitoring network is built to achieve real-time perception of equipment failures and path conflicts; based on the warehouse topology, multi-priority backup paths are pre-generated to form an intelligent emergency response mechanism; edge computing nodes are used to achieve local decision-making and path switching, eliminating remote communication delays; and finally, high-precision motion control algorithms are used to ensure robot execution efficiency. This solution can effectively deal with dynamic scenarios such as freight elevator failures and robot congestion, reduce manual intervention, improve cargo warehousing efficiency, and solve technical problems such as poor equipment coordination and delayed response in traditional warehouse systems. Figure 2 .

[0032] Furthermore, the real-time data interface includes an industrial communication protocol interface for data interaction with the freight elevator controller and the robot controller, and the dynamic event information includes freight elevator fault signals and robot position conflict detection data.

[0033] In this embodiment, the intelligent warehousing system's industrial communication protocol interface serves as a protocol conversion hub between heterogeneous devices, enabling command exchange and status feedback between the freight elevator and robot control units. By being compatible with multiple industrial communication protocols such as OPC UA, Modbus, and CANopen, a standardized data exchange channel is established, enabling real-time data exchange between the freight elevator controller and the robot controller through a unified interface.

[0034] Furthermore, the freight elevator fault signal obtains the operating status data of the freight elevator PLC through the industrial Ethernet interface, and the operating status data includes the freight elevator position, load weight, and door switch status.

[0035] In this embodiment, the freight elevator controller uses a programmable logic controller (PLC). It interacts with PLCs from different manufacturers via industrial communication protocols, directly reading operational parameters such as the elevator's floor position and load weight. The system converts data formats using standard protocols for different PLC brands, ensuring protocol compatibility and system identity across different manufacturers' freight elevators. The system collects vibration signals from the elevator motor in real time via an industrial Ethernet interface and reads data from the PLC's internal fault registers. If an abnormal elevator operation is detected, the system immediately transmits the fault information to the system's core processing unit, providing data support for subsequent fault diagnosis and resolution.

[0036] Regarding robot positioning and communication, the system uses high-precision positioning technology to acquire the robot's three-dimensional coordinate data, packaging and transmitting this data using standardized protocols. Furthermore, it communicates with the robot controller via an industrial fieldbus protocol, acquiring real-time status parameters such as the robot's power level and operating mode, enabling comprehensive monitoring of the robot's operational status. When the system needs to issue new path instructions, data is transmitted via the IoT communication protocol, which features a retransmission mechanism to ensure reliable transmission of instructions even in fluctuating network environments.

[0037] To address the heterogeneous protocols between freight elevators and robots, the system deploys protocol conversion middleware. This uses an industrial protocol client to discover freight elevator nodes and a protocol converter to convert data into a unified format. All devices implement clock calibration via a time synchronization protocol to avoid control errors caused by time deviations. To ensure communication reliability, the system employs a dual-link redundant design, with each device connected to both industrial Ethernet and wireless LAN. The system periodically sends heartbeat packets to check device online status. If no response is received continuously, an early warning mechanism is triggered to ensure timely resolution of communication anomalies. By refreshing freight elevator and robot operating data in real time, the system dynamically monitors device status, providing an accurate data foundation for path planning and troubleshooting.

[0038] Furthermore, the robot position conflict detection data includes:

[0039] S201. Acquire three-dimensional coordinate data of each storage robot in real time through a positioning system, wherein the positioning accuracy of the positioning system meets the positioning error requirements preset by the system;

[0040] S202. Acquire robot position conflict data through a deployed laser radar sensor, wherein the scanning frequency and angular resolution of the laser radar meet the detection accuracy requirements preset by the system.

[0041] In this embodiment, the system configures a three-dimensional spatial positioning system for each warehouse robot, dynamically tracking its spatial position in the warehouse environment through real-time coordinate acquisition technology. The positioning system uses ultra-wideband (UWB) technology or laser SLAM technology to accurately output the robot's position data (X, Y, Z) in a three-dimensional coordinate system. For example, the coordinates in the plane coordinate system of the third floor of the warehouse are (10m, 5m, 2m). The system presets a positioning error threshold to ensure that the positioning data accuracy meets the requirements of the path planning algorithm, avoiding path conflicts or operation interruptions due to positioning deviations.

[0042] The system also deploys lidar sensors to build an environmental perception system, which scans the surrounding environment 360° at a fixed frequency and angular resolution. Using point cloud data processing technology, it can not only identify static obstacles (such as walls and shelves) but also track the spatial position of dynamic targets (such as other robots) in real time. The positioning system and lidar form the robot's environmental perception unit: the former provides the robot's precise coordinates, while the latter generates a real-time point cloud map of the surrounding environment.

[0043] When the LiDAR detects another robot entering the operating area, the system combines positioning data with multi-sensor data fusion processing and uses the Kalman filter algorithm to calculate relative speed and minimum approach distance. For example, if robot M is detected approaching robot N at a speed of 0.8m / s and the two robots are predicted to enter the safety threshold within 2 seconds, the system immediately triggers the path planning mechanism. This multi-sensor fusion architecture effectively reduces the probability of misjudgment by a single sensor, improves the accuracy of conflict detection through data complementarity, and ensures safe operation of robots in high-density operating environments.

[0044] Furthermore, after obtaining the robot position conflict data through the positioning system and the laser radar sensor, the method further includes:

[0045] S301: performing timestamp synchronization on acquired multi-source heterogeneous data, where the multi-source heterogeneous data includes data acquired by the positioning system and data acquired by the lidar, and an error in the timestamp synchronization meets a synchronization accuracy requirement preset by the system;

[0046] S302, using a data fusion algorithm to process the robot position data to eliminate random errors of the positioning system;

[0047] S303: Calculate conflict characteristic parameters such as relative speed and minimum approach distance of adjacent robots based on the processed position data.

[0048] In this embodiment, the system deploys a time synchronization mechanism for heterogeneous devices. Clock deviations exist between the raw data from the positioning system and the lidar. The system calibrates timestamps using technologies such as the Network Time Protocol (NTP) to ensure consistent time bases across multiple data sources. The system sets strict synchronization accuracy standards, such as limiting time deviations to less than 10 microseconds, to avoid position calculation errors caused by time deviations in high-speed motion scenarios. Furthermore, the system uses a data fusion algorithm to reduce noise in position data. UWB positioning systems may generate random errors due to multipath effects, and lidars may also experience data offsets due to interference from ambient light. Using algorithms such as Kalman filtering, combining the long-term positioning stability of UWB with the real-time, high-precision characteristics of lidar, a data complementarity correction model is constructed to achieve dynamic calibration of positioning data, ultimately outputting high-precision robot pose data. Based on the fused position data, the system constructs a collision risk assessment model. By calculating characteristic parameters such as the relative speed and minimum approach distance of adjacent robots, a risk quantification indicator system is established. For example, when the relative speed of two robots reaches 1 m / s and the predicted minimum approach distance is less than 0.8 meters, the system determines a high-risk collision state. The model uses a real-time risk assessment mechanism to achieve early warning of potential collision risks and autonomous planning of obstacle avoidance paths. Its decision-making logic is equivalent to the danger prediction mechanism in intelligent transportation systems.

[0049] Furthermore, the dynamic path adaptive scheduling mechanism based on the path planning algorithm includes:

[0050] S401. Construct a basic path planning model based on a graph search algorithm;

[0051] S402, introducing time window constraints and building a multi-robot path conflict avoidance model;

[0052] S403: Optimize and train the path planning model based on a machine learning algorithm.

[0053] In this embodiment, the graph theory method is used to abstract the physical space of the warehouse into a mathematical model. First, a node-edge network graph (G = (V, E)) is created based on the warehouse layout, where: the node set V includes key location points such as shelf locations, lane intersections, elevator entrances, charging areas, and each node is marked with three-dimensional coordinates (x, y, z); the edge set represents the passable paths between nodes, and each edge is assigned a distance weight and a travel cost. The A* algorithm is used as the core path search algorithm. The distance from node n to the target node is evaluated by defining the heuristic function h(n), and the total cost f(n) = g(n) + h(n) is calculated in combination with the actual cost g(n). For example: the heuristic function uses Euclidean distance to ensure that the estimated distance does not exceed the actual shortest distance; dynamic weight adjustment is to dynamically adjust the weight of the edge according to real-time congestion data during runtime.

[0054] At the same time, in order to solve the problem of multi-robot path conflict, the time dimension constraint is introduced to build a spatiotemporal path planning model. When the time window is used, a time interval [t s ,t e ], representing the time period that a robot can occupy. If two robots enter adjacent nodes within the same time window, a potential collision occurs. If the robot's arrival time at the elevator entrance does not match the elevator's operating cycle, a waiting constraint is triggered. A time expansion graph is used to embed the time dimension into the basic network graph, with each node expanded into multiple copies at different time steps. A resource-constrained project scheduling algorithm is applied, and the optimal time window allocation solution is solved using the branch and bound method. Here, x(v,e,t) is a binary decision variable indicating whether the edge (v,e) is occupied at time t, and t is the travel cost.

[0055] Furthermore, to improve the efficiency of path planning in dynamic environments, a deep reinforcement learning optimization model is used. The state space S contains the robot's position, speed, task priority, remaining battery power, global path congestion status, etc.; the action space A is the set of discrete actions to select the next node, such as {A1 = go to node N1, A2 = go to node N2, ...}; the reward function R is:

[0056] R=w1·Complete the task+w2·Avoid conflict-w3·Path length-w4·Waiting time

[0057] Among them, w1=100, w2=50, w3=0.1, and w4=0.2 are weight coefficients.

[0058] At the same time, the proximal policy optimization algorithm PPO is used to train the intelligent agent, and a multi-layer perceptron is used as the policy network; an experience replay buffer is designed to store the state-action-reward sequence to improve the efficiency of sample utilization; a classroom learning mechanism is introduced, gradually transitioning from a simple scenario with 5 robots to a complex scenario with 50 robots. 6 The training is conducted in 50 time steps, with each episode containing 50 time steps. A parameter sharing mechanism is adopted to enable the agent to generalize to robot clusters of different sizes. Verification is performed in a real environment every 1000 steps, and hyperparameters are adjusted based on the verification results.

[0059] Furthermore, the preset backup path set includes:

[0060] S501. Pre-generate at least three backup paths of different priorities based on the three-dimensional map of the warehouse;

[0061] S502: The path priority is determined by comprehensively determining at least two of the following: path length, energy consumption coefficient, device usage frequency, and congestion probability;

[0062] S503: When the primary path is occupied or fails, select a backup path in order of priority.

[0063] In this embodiment, based on a three-dimensional map with a warehouse modeling accuracy of ±5cm, an improved A* algorithm is used to pre-generate at least three backup paths. First, the warehouse space is discretized into a 0.5m×0.5m×2m three-dimensional grid, with each grid node containing attributes such as accessibility and load-bearing restrictions. Then, with the starting point and target point as endpoints, differentiated paths are generated by adjusting the heuristic function weights and temporary constraints. For example, the first path focuses on the shortest distance, and the heuristic function uses Euclidean distance; the second path avoids the frequently used No. 1 freight elevator, forcing the algorithm to select No. 2 freight elevator; and the third path restricts areas with aisle widths less than 2m to generate a detour. At least 30% of the nodes on the three paths do not overlap in spatial direction, ensuring the feasibility of alternatives in the event of a failure on the main path.

[0064] The priority of backup paths is determined by combining at least two of four metrics: path length, energy consumption coefficient, equipment usage frequency, and congestion probability. Path length is based on the primary path length, with a deduction of 10 points for every 10% increase, and a bonus of 5 points for any decrease. The energy consumption coefficient is calculated based on the robot's trajectory's acceleration and number of turns. The energy consumption coefficient for smooth paths is set at 1.0, and increases by 0.2 for each sharp turn. Equipment usage frequency is calculated by counting the number of times the freight elevators and lanes involved in the path are used over the past 24 hours. If the number of equipment usage exceeds the average by 20%, a 15-point deduction is applied to the corresponding path. A sliding window method is used based on historical traffic data and real-time sensor data to calculate the congestion probability of the path. If the probability exceeds 30%, a 20-point deduction is applied. For example, a backup path with a length of 110% of the primary path (a 10-point deduction), an energy consumption coefficient of 1.2 (a 4-point deduction), and a freight elevator usage frequency below the average (a bonus of 8 points) would result in a combined score of -6 points, placing it in second place.

[0065] Furthermore, when the system detects that a path is occupied, such as a freight elevator failure or robot congestion, it automatically switches to the preset backup path in priority order. The detection mechanism includes: judging the elevator operation status through the freight elevator PLC data, calculating the lane congestion density using the lidar point cloud data, and calculating the lane congestion density of more than 0.5 units / m 2 Determined to be congested. The switching logic is as follows: First, the planned path with the highest priority is activated. If new congestion occurs on this path during switching, and the real-time congestion probability is greater than 50, the secondary path evaluation is immediately triggered and the priority is dynamically adjusted. For example, if the route fails due to a failure of freight elevator No. 1, the system will prioritize the use of backup path A that bypasses freight elevator No. 1. If path A detects that the queue time for freight elevator No. 2 exceeds 3 minutes during switching, the priority of path A will be automatically downgraded, and backup path B, which requires a detour but has an idle freight elevator, will be activated to ensure that the path switching response time is controlled within 200ms.

[0066] Furthermore, the edge computing unit executes a backup path calculation strategy, including:

[0067] S601, a multi-core processor based on an edge computing unit calculates multiple candidate paths in parallel;

[0068] S602: Using a heuristic search algorithm to complete the optimal path selection within a system preset time threshold;

[0069] S603: When computing resources are insufficient, some computing tasks are offloaded to the cloud server.

[0070] In this embodiment, the edge computing unit uses the Intel Xeon D-2146NT multi-core processor to build a parallel computing architecture and perform hardware acceleration on the backup path calculation task. The specific implementation method is: the three-dimensional map of the warehouse is divided into multiple computing subgraphs according to the aisles and shelf areas, and each subgraph is assigned an independent thread for path search. For example, when generating three backup paths, the system simultaneously starts three threads to execute the A* algorithm under different heuristic function configurations: thread 1 uses the shortest distance heuristic, thread 2 uses the lowest energy consumption heuristic, and thread 3 uses the least device dependency heuristic. The hyperthreading technology of the multi-core processor enables each physical core to process two threads in parallel, significantly improving computing efficiency and completing the preliminary generation of three candidate paths within 500ms.

[0071] At the same time, in order to ensure that the path calculation is completed within the system's preset time threshold of 500ms, this embodiment adopts an improved version of the iterative deepening A* (IDA*) algorithm. The specific strategy is as follows: first, the initial search depth is set to 1.2 times the estimated shortest path length, and a limited search is performed within this depth; if no feasible path is found, the depth is increased by 0.2 times and the search continues until the time threshold is reached. At the same time, a pruning strategy is introduced to reduce invalid calculations: when the current cost value of a path exceeds 150% of the optimal value of the found path, the path search is terminated. For example, when searching for the third backup path, the system detects that the calculation time of a branch path has reached 350ms and the cost value continues to rise, then the branch is immediately abandoned and turned to other possible paths. Through this mechanism, the system can complete at least 80% of the path evaluation within the time threshold, ensuring the response speed of critical tasks.

[0072] Furthermore, when the load of the edge computing unit exceeds 70%, the system automatically triggers the computation offloading mechanism. The specific process is: tasks with high computational complexity, such as global path optimization and long-term prediction simulation, are encapsulated into containerized microservices and transmitted to the cloud server for processing via the 5G network. For example, when more than 5 robots request a backup path calculation strategy at the same time, the edge unit offloads secondary tasks, such as the fine optimization of the third backup path, to the cloud, while retaining core tasks such as the rapid generation of the main path and the first backup path. The cloud server uses AWS EC2 P3 instances to accelerate computing, and the processing results are fed back to the edge unit via the MQTT protocol. The delay of the entire offloading process is controlled within 150ms. Through this collaborative mechanism, the system can still maintain a path calculation success rate of more than 95% under peak load, ensuring the stability of the warehousing and logistics system.

[0073] Furthermore, controlling the storage robot to perform warehousing operations according to the planned path includes:

[0074] S701, generating robot motion control instructions based on the path planning results, wherein the instructions include a speed curve, a steering angle, and acceleration parameters;

[0075] S702, transmitting the control instructions to the storage robot in real time via the industrial wireless communication network;

[0076] S703: Using a closed-loop control algorithm to control the robot's motion trajectory, the trajectory tracking error meets the accuracy requirements preset by the system.

[0077] In this embodiment, the motion control instructions generated based on the path planning results contain refined parameters: speed curve, steering angle, and acceleration parameters. The speed curve is dynamically adjusted according to the path curvature. The straight section is set to a maximum speed of 1.5m / s, and when the turning radius R < 2m, it is reduced to 0.8m / s to ensure that the centrifugal force does not exceed 0.3g; the steering angle is calculated based on the path tangent direction, and an S-shaped acceleration and deceleration strategy is adopted. When the steering angle is > 30°, a pre-deceleration section is added to ensure smooth steering; the starting acceleration is set to 0.5m / s 2 , the braking acceleration is set to 0.8m / s 2 To prevent cargo from shaking (vibration threshold ≤ 0.5g). For cross-floor operations, the command also includes elevator docking parameters: decelerate to 0.2m / s 3m in advance, and control the position error within ±5mm when aligning with the elevator door.

[0078] The system also uses a time synchronization protocol to ensure that the clock deviation between command transmission and reception is less than 1μs, avoiding control errors caused by time asynchrony. Control commands are transmitted via an industrial wireless communication network that integrates Wi-Fi 6 and 5G.

[0079] Furthermore, a dual closed-loop control system is constructed using a PID+feedforward control algorithm. The position loop uses the path point as the target value, with a sampling period of 10ms, a proportional coefficient Kp=1.2, an integral coefficient Ki=0.05, and a differential coefficient Kd=0.1 to eliminate static errors; the speed loop uses the speed curve as the target value, with a sampling period of 5ms, Kp=0.8, Ki=0.03, and Kd=0.05 to suppress speed fluctuations; the feedforward compensation outputs the steering torque in advance according to the path curvature, compensates for inertia delay, and improves dynamic response. The system obtains the robot position in real time through laser SLAM, compares it with the command path, and triggers the correction mechanism when the trajectory error exceeds ±5cm: error compensation is completed within 500ms, and the correction acceleration is ≤0.3m / s 2 , ensuring the stability of the cargo. Actual measurement data shows that when fully loaded with 500kg, the trajectory error of the straight section is ≤±2cm, and the error of the turning section is ≤±3cm, meeting the system's preset accuracy requirement of ±5cm.

[0080] In this embodiment, if Figure 3 , provides an automated goods warehousing system for intelligent warehousing, which is used to implement the automated goods warehousing method for intelligent warehousing, including:

[0081] The data perception module is used to obtain elevator fault signals and robot position conflict data through sensors;

[0082] A path planning module having a built-in dynamic path adaptive scheduling mechanism and a backup path set, wherein the backup path set includes at least one backup path;

[0083] An edge computing unit, configured to complete a backup path calculation strategy within a preset time threshold range when a dynamic event is detected;

[0084] The execution control module is used to control the storage robot to perform warehousing operations according to the planned path;

[0085] Among them, the data perception module, path planning module, edge computing unit and execution control module work together to achieve real-time path optimization under dynamic events.

[0086] In this embodiment, the data perception module serves as the information collection center of the system, and builds an all-round monitoring network through multiple types of sensors: vibration, temperature, and current sensors are deployed in key components of the freight elevator to perceive abnormal motor operation, door machine failure and other conditions in real time; data such as floor position and door opening and closing status are obtained from the freight elevator PLC through industrial communication protocols to achieve digital monitoring of the entire life cycle of the freight elevator; high-precision positioning technology is used to obtain the three-dimensional coordinates of the warehouse robot in real time, and combined with laser radars deployed in lanes and intersections, the surrounding environment is continuously scanned to identify the robot's position, movement trajectory and potential collision risks; infrared sensors, proximity switches and other equipment are used to monitor the relative distance between the robot and shelves and obstacles to ensure safe operation. At the same time, the module performs multi-layer processing on the collected raw data, converting it into effective event information: It uses filtering algorithms to eliminate noise interference in sensor data, and uses timestamp synchronization technology to ensure the temporal consistency of multi-source data to avoid misjudgments due to time deviations; it extracts key features from vibration, current, and other data, and combines them with machine learning algorithms to establish an equipment health model to identify fault characteristics such as abnormal vibration and current overload of freight elevators; based on the robot's motion trajectory and speed data, it uses a collision prediction algorithm to determine potential conflict risks; and it classifies detected problems according to the degree of abnormality (such as minor abnormality and emergency failure). When the preset threshold is reached, dynamic event information (such as "freight elevator motor abnormality" and "robot path conflict") is immediately generated and pushed to the system's core processing unit. Through the above design, the data perception module can quickly and accurately capture dynamic changes in the warehouse environment, providing a reliable information foundation for subsequent path planning and task scheduling, effectively improving the system's intelligent decision-making capabilities and emergency response efficiency.

[0087] At the same time, the path planning module, as the core decision-making unit of the system, integrates dynamic path planning and backup path management functions: based on the three-dimensional map of the warehouse, a graph search algorithm is used to pre-generate multiple basic paths from the entrance to each target shelf and store them in the path database. Each path contains parameters such as node sequence, turning point coordinates, and estimated travel time, providing basic data support for subsequent planning; built-in time window constraints and priority evaluation mechanisms, when the data perception module detects dynamic events such as freight elevator failures and robot conflicts, the system immediately triggers path planning, and assigns a priority to each backup path by comprehensively evaluating factors such as path length, equipment usage frequency, and congestion probability, and automatically selects the optimal path according to the priority order; combined with historical operation data and real-time status feedback, the path planning module can dynamically adjust the path weight and priority strategy. For example, if a path is frequently congested, its priority will be lowered; if the backup path performs well in actual use, the weight will be increased to achieve continuous optimization of the path planning strategy.

[0088] Furthermore, the edge computing unit, as the real-time computing core of the system, is responsible for rapid path decision-making and task offloading: based on the multi-core processor architecture, multiple candidate paths are calculated in parallel, significantly shortening the path planning time. By optimizing the algorithm and collaborating with the hardware, it ensures that the calculation and screening of alternative paths are completed within the time threshold preset by the system, meeting the strict real-time requirements of the warehousing scenario; when the computing resources are sufficient, the edge computing unit independently completes the path planning task. If the load is too high, non-urgent computing tasks (such as path fine optimization) are automatically offloaded to the cloud server, realizing collaborative processing between the edge and the cloud, and avoiding response delays due to insufficient resources; the calculation results of commonly used paths are cached in local storage. When the same task is requested again, the cached data is directly called to reduce repeated calculations, further improving the system response speed, and ensuring that the robot can quickly obtain new path instructions under dynamic events.

[0089] Furthermore, the execution control module, as the terminal execution unit of the system, is responsible for converting the path planning results into the actual actions of the robot: based on the curvature, distance and other information of the planned path, it generates motion control instructions containing speed curves, steering angles, and acceleration parameters. For example, it sets the maximum safe speed in the straight section, automatically reduces the speed and adjusts the steering angle in the turning area to ensure smooth operation of the robot; transmits the control instructions to the warehouse robot in real time through the industrial wireless communication network (such as 5G, Wi-Fi 6), and establishes a two-way communication mechanism to receive real-time status feedback such as the robot's position and speed to realize dynamic monitoring of the robot's operating status; adopts the PID closed-loop control algorithm to compare the actual motion trajectory of the robot with the planned path in real time. When the trajectory error exceeds the preset range, the control parameters are automatically adjusted to ensure that the robot strictly performs the warehousing operation according to the planned path, and finally achieves that the trajectory tracking error meets the system accuracy requirements, thereby ensuring the accuracy and safety of cargo transportation.

[0090] In summary, the present invention has constructed a complete and efficient intelligent warehousing cargo automation system through the coordinated operation of the data perception module, the path planning module, the edge computing unit and the execution control module. As the front-end information acquisition unit of the system, the data perception module can accurately capture dynamic events such as freight elevator failures and robot path conflicts; the path planning module presets alternative paths and implements dynamic optimization based on a multi-dimensional evaluation system, constituting the decision-making core of the system; the edge computing unit ensures that the path planning is completed within the system preset time threshold through localized parallel computing, achieving rapid response; the execution control module ensures that the warehouse robot accurately performs operations according to the planned path through a high-precision motion control algorithm. The system breaks through the technical bottleneck of frequent manual intervention and delayed response in traditional warehousing, and realizes the automation and intelligence of the entire process from anomaly detection, path planning to task execution. It has been verified in practice that the system can significantly improve the efficiency of warehousing logistics operations, reduce equipment failure rates, and provide innovative technical solutions for the efficient and stable operation of modern intelligent warehousing.

[0091] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. A method for automated cargo entry for intelligent warehousing, wherein the method is applied to an automated warehousing system, which is used to control warehousing robots and includes multiple edge computing units, characterized in that: The automated warehousing method comprises: Obtain dynamic event information in the cargo entry scenario through a real-time data interface; When a dynamic event is detected, at least one edge computing unit is selected to jointly generate at least one backup path for the warehouse robot based on the dynamic event information, the location information of the warehouse robot, and the surrounding image information collected by the warehouse robot, wherein the duration of the joint processing by the selected edge computing units is shorter than the system preset duration; Control the storage robot to perform warehousing operations according to one of the alternative paths.

2. The method according to claim 1, characterized in that The real-time data interface includes an industrial communication protocol interface for data interaction with the freight elevator controller and the robot controller. The dynamic event information includes freight elevator fault signals and robot position conflict detection data.

3. The method according to claim 2, characterized in that The freight elevator fault signal further includes: The operating status data of the freight elevator PLC is obtained through the industrial Ethernet interface. The operating status data includes the freight elevator position, load weight, and door switch status.

4. The method according to claim 2, characterized in that The robot position conflict detection data further includes: The three-dimensional coordinate data of each storage robot is obtained in real time through the positioning system, and the positioning accuracy of the positioning system meets the positioning error requirements preset by the system; The robot position conflict data is obtained by deploying a lidar sensor, the scanning frequency and angular resolution of which meet the detection accuracy requirements preset by the system.

5. The method according to claim 4, characterized in that After obtaining the robot position conflict data through the positioning system and the laser radar sensor, the method further includes: Performing timestamp synchronization on the acquired multi-source heterogeneous data, the multi-source heterogeneous data including data acquired by the positioning system and data acquired by the lidar, wherein an error in the timestamp synchronization meets a synchronization accuracy requirement preset by the system; Use data fusion algorithm to process robot position data to eliminate random errors in the positioning system; Based on the processed position data, the conflict characteristic parameters such as the relative speed and minimum approach distance of adjacent robots are calculated.

6. The method according to claim 1, characterized in that The dynamic path adaptive scheduling mechanism based on the path planning algorithm further includes: Build a basic path planning model based on graph search algorithm; Introducing time window constraints to build a multi-robot path conflict avoidance model; Optimize and train the path planning model based on machine learning algorithms.

7. The method according to claim 6, characterized in that The preset backup path set further includes: Pre-generate at least three backup routes with different priorities based on the warehouse's three-dimensional map; The path priority is determined by comprehensively determining at least two of the path length, energy consumption coefficient, device usage frequency, and congestion probability; When the primary path is occupied or fails, a backup path is selected in order of priority.

8. The method according to claim 1, characterized in that The edge computing unit executes the backup path computing strategy, further comprising: The multi-core processor based on the edge computing unit calculates multiple candidate paths in parallel; Use heuristic search algorithms to complete the optimal path screening within the system preset time threshold; When computing resources are insufficient, some computing tasks are offloaded to cloud servers.

9. The method according to claim 1, characterized in that The controlling the storage robot to perform the warehousing operation according to the planned path further includes: Generate robot motion control instructions based on the path planning results, the instructions including speed curve, steering angle, and acceleration parameters; Transmit control instructions to warehouse robots in real time through industrial wireless communication networks; A closed-loop control algorithm is used to control the robot's motion trajectory, and the trajectory tracking error meets the system's preset accuracy requirements.

10. An automated cargo storage system for intelligent warehousing, characterized in that: include: The data perception module is used to obtain elevator fault signals and robot position conflict data through sensors; A path planning module having a built-in dynamic path adaptive scheduling mechanism and a backup path set, wherein the backup path set includes at least one backup path; An edge computing unit, configured to complete a backup path calculation strategy within a preset time threshold range when a dynamic event is detected; The execution control module is used to control the storage robot to perform warehousing operations according to the planned path; Among them, the data perception module, path planning module, edge computing unit and execution control module work together to achieve real-time path optimization under dynamic events.

Images