Intelligent logistics transfer device

Abstract

The utility model relates to the technical field of intelligent logistics and cooperative scheduling control, and particularly provides an intelligent logistics transfer device. The device comprises a transferring device, a swing arm structure, a grabbing mechanism and an intelligent cooperative scheduling module. The system integrally supports task state feedback, path obstacle avoidance and cluster scheduling execution, is suitable for flexible storage, intelligent carrying and multi-target material circulation scenes, and has the remarkable technical advantages of intelligent control, path optimization, task collaboration and the like.

Description

Technical Field

[0001] This utility model relates to the field of intelligent logistics and task scheduling technology, and in particular to an intelligent logistics transfer device. Background Technology

[0002] In modern intelligent manufacturing, warehousing and logistics, and urban distribution, the precise transfer and efficient scheduling of materials have become key factors affecting the overall efficiency of the system. Traditional logistics transfer equipment mostly relies on a single robotic arm or fixed track structure. Although it can complete basic grabbing, handling, and unloading operations, it has obvious limitations when facing dynamic multi-tasking environments.

[0003] In existing technologies, common logistics transfer devices often use preset routes or manual control methods, which cannot be dynamically adjusted according to real-time task requirements or operational status. This can easily lead to path conflicts, resource waste, or task scheduling delays in scenarios with dense tasks or complex spaces. At the same time, most existing systems lack a unified scheduling center, with each device operating independently without a coordination mechanism, making it impossible to perform cluster-based optimization and load balancing management.

[0004] Furthermore, most current logistics scheduling systems still rely on static rule engines, lacking learning-based path planning capabilities and unable to automatically optimize scheduling strategies based on historical data or operational feedback. While some devices incorporate sensor terminals to collect status data, a complete data loop has not yet been formed in the processes of task allocation, path generation, and multi-device collaboration. Faced with highly dynamic and unstructured task environments, traditional scheduling models struggle to achieve high-precision, high-response intelligent control.

[0005] Therefore, there is an urgent need to develop an intelligent logistics transfer device with intelligent path planning capabilities, adaptive task allocation, and support for collaborative scheduling of multiple devices. This device should integrate reinforcement learning algorithms, task scheduling strategies, and sensor feedback mechanisms to achieve intelligent linkage throughout the entire process from order data input to actual path execution, thereby significantly improving the operational efficiency, flexibility, and scheduling intelligence level of the logistics system. Utility Model Content

[0006] The purpose of this invention is to provide an intelligent logistics transfer device. This device integrates a transfer structure unit, a multi-degree-of-freedom grasping component, and an intelligent collaborative scheduling module, enabling automatic identification, grasping, transfer, and unloading of materials. The system combines a task allocation submodule and a path planning submodule to support multi-device collaborative execution, path obstacle avoidance, and optimized task scheduling, making it particularly suitable for warehousing and logistics environments with multiple orders, high concurrency, and dynamic path scenarios.

[0007] To achieve the above objectives, this utility model provides an intelligent logistics transfer device, comprising:

[0008] The transfer device specifically includes: a transport vehicle unit, including a loading section and a column extending upward from the loading section; a swing arm unit, one end of which is connected to the column, the swing arm unit including an inclined arm and a proximal arm, the inclined arm being rotatably connected to the column, and the inclined arm and the proximal arm being rotatably connected along a longitudinal axis; a sensor group; a gripping unit connected to the distal end of the proximal arm; and an intelligent collaborative scheduling module, the intelligent collaborative scheduling module controlling the movement paths of the swing arm unit and the gripping unit according to task instructions.

[0009] The gripping unit specifically includes an unloading action; the unloading action is: the swing arm unit moves the transported object from the location of the transport vehicle unit to the target unloading position, and disconnects the object from the transported object at the target unloading position.

[0010] The gripping unit specifically includes a gripping action; the gripping action is: the gripping unit grips the transfer object at the target location, and the swing arm unit drives it to be transferred to the transport vehicle unit to complete the placement operation.

[0011] The intelligent collaborative scheduling module includes a task allocation submodule, whose input parameters include: the number of sensor terminal components, the list of tasks to be assigned, the learning rate, the discount factor, the exploration rate and the exploration rate decay coefficient, and dynamically allocates the task objectives of each terminal component.

[0012] The task allocation submodule and the path planning submodule communicate via a data stream and module interface. The task allocation submodule transmits the task allocation results of each sensing terminal component to the path planning submodule.

[0013] The intelligent collaborative scheduling module further includes a path planning submodule, which generates the optimal task path based on the Q-learning algorithm and transmits the path instructions to each transfer device.

[0014] The intelligent collaborative scheduling module receives external order information as a data source and generates control commands based on the scheduling results to drive the transfer device to complete the task execution; multiple transfer devices form a collaborative cluster, which is controlled by a unified intelligent collaborative scheduling module to achieve multi-task synchronous allocation and path conflict de-conflict optimization.

[0015] Furthermore, the task allocation submodule in the intelligent collaborative scheduling module is equipped with a strategy learning interface, which supports dynamic adjustment of the learning rate, discount factor and exploration rate decay mechanism through long-term running data, thereby adaptively optimizing the allocation strategy according to task density and resource utilization, and improving task execution efficiency and device response capability.

[0016] Furthermore, the path planning submodule integrates a reinforcement learning algorithm based on Q-learning. By constructing a state-action-reward model, it continuously iterates the path strategy during each task execution, adapting to scenarios of multi-terminal collaboration, multi-path interference, and dynamic target adjustment, and supporting local obstacle avoidance and target replanning capabilities.

[0017] Furthermore, the data flow and module interface in the intelligent collaborative scheduling module are efficiently connected based on standard communication protocols, and have the ability to transmit asynchronous data between modules, error retry mechanism and task status feedback, ensuring the stability of task instruction transmission and the synchronization of information within the system.

[0018] Furthermore, multiple transfer devices are coordinated by a unified scheduling center through a collaborative cluster approach. The scheduling center allocates resources and plans routes based on global order information, supports horizontal expansion, and is suitable for automatic transfer scheduling in large-scale distributed logistics warehousing systems.

[0019] Furthermore, the gripping unit (130) supports gripping actions for various types of workpieces or goods. Its structure can be switched to a gripper type, adsorption type or hook type actuator component according to the shape of the goods, adapting to the handling needs of different materials and improving the system's flexible processing capability.

[0020] Furthermore, the swing arm unit (120) supports a six-degree-of-freedom motion structure and is equipped with an angle encoder and a motion control driver to provide real-time feedback of arm posture information. Combined with the scheduling module, it performs posture correction and path refinement control to achieve high-precision position positioning and trajectory tracking.

[0021] This utility model has the following beneficial effects:

[0022] By integrating gripping, swing arm, and intelligent scheduling control modules into the transfer device, automated transfer operations of materials in different locations and of different types can be achieved, enhancing the device's flexibility and adaptability in complex operating scenarios. The gripping unit has bidirectional operation capabilities for unloading and gripping, and combined with the multi-degree-of-freedom swing arm mechanism, it can achieve precise positioning, spatial obstacle avoidance, and continuous operation, effectively reducing human intervention and operational errors.

[0023] By introducing an intelligent collaborative scheduling module, the system supports dynamic task allocation and real-time path optimization across multiple terminals. The task allocation submodule adaptively adjusts tasks based on learned parameters, while the path planning submodule uses a Q-learning reinforcement algorithm to continuously iterate the optimal path strategy, enabling the system to continuously optimize. This mechanism is suitable for warehousing or delivery environments with high task concurrency and frequent path conflicts, significantly improving transfer efficiency and system throughput.

[0024] Multiple transfer devices can form a collaborative cluster, performing synchronous scheduling and coordinated operations under the unified control of the scheduling module. It possesses distributed control and centralized strategy feedback capabilities, supporting large-scale material handling operations. The system interfaces with the data flow module through a standard interface, ensuring rapid transmission and execution of task instructions, sensor data, and status feedback among sub-modules, achieving closed-loop information control and real-time status awareness.

[0025] Furthermore, the device boasts a compact structure and high mobility, making it suitable for various warehousing layouts and logistics transfer environments. The system can integrate order source data as scheduling input, combining real-time path calculation and execution command issuance to construct a complete automated processing loop from order initiation to material placement, demonstrating excellent practicality, scalability, and a solid foundation for intelligent upgrades. This device is particularly suitable for medium to large-sized logistics warehouses, automated distribution centers, or industrial material distribution scenarios, possessing broad application prospects and promotional value. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments or examples of this utility model, the drawings used in the embodiments or examples will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this utility model. For those skilled in the art, other drawings can be obtained according to these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of the structure of an intelligent logistics transfer device;

[0028] Figure 2 This is a control flowchart for an intelligent logistics transfer device processing system.

[0029] Figure 3 This is the control flowchart of the Q-learning algorithm within the intelligent collaborative scheduling module of this system.

[0030] Explanation of reference numerals in the attached drawings: 100-Transfer device; 101-Transfer object; 110-Transfer vehicle unit; 111-Loading section; 112-Column section; 113-Sensor group; 120-Swing arm unit; 121-Tilt arm; 122-Proximal arm; 130-Grip unit. Detailed Implementation

[0031] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0032] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0033] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0034] See Figure 1 This is a schematic diagram of the structure of an intelligent logistics transfer device. In this embodiment, the device includes a transfer device 100, which consists of a transport vehicle unit 110, a swing arm unit 120, a gripping unit 130, a sensor group 113, and an intelligent collaborative scheduling module. It is used to complete the entire process of automatic object grabbing, path transfer, and target placement under intelligent control commands.

[0035] The transport vehicle unit 110 includes a loading section 111 and a column section 112 extending vertically from the loading section 111. Four sets of omnidirectional wheel assemblies are located below the loading section 111, and an integrated shock-absorbing structure is provided, enabling the device to travel smoothly in any direction on the warehouse floor. The column section 112 is a load-bearing connecting component, and a swing arm unit 120 is mounted on its top.

[0036] The swing arm unit 120 consists of a tilting arm 121 and a proximal arm 122. One end of the tilting arm 121 is connected to the column 112 via a rotating mechanism, allowing it to rotate approximately 180° horizontally. The proximal arm 122 is rotatably connected to the tilting arm 121 via a joint axis, enabling the gripping action to be adjusted in both height and angle. This multi-joint design allows the gripping unit 130 to be precisely aligned within a spatial range.

[0037] The gripping unit 130 is located at the distal end of the proximal arm 122 and is a universal electric gripper structure with a gripper opening width ranging from 0 to 240 mm and a maximum load capacity of approximately 6.5 kg. The gripper is equipped with force feedback and anti-pinch error detection mechanisms, which can identify whether the target object has been successfully gripped and promptly interrupt the operation in case of error to ensure operational safety.

[0038] The intelligent collaborative scheduling module is deployed in the control motherboard and includes a task allocation submodule and a path planning submodule. The task allocation submodule dynamically allocates and schedules tasks based on parameters such as the number of terminals, the task list, and the environmental status; the path planning submodule uses the Q-learning algorithm for path planning. The Q-learning process uses a warehouse plane grid (40×40) as the state space, with each movement being an action, and the goal is to complete loading and unloading using the shortest path while avoiding obstacles.

[0039] The following is a complete operational example:

[0040] The system receives an instruction from the warehouse dispatch center to move item 101 from storage area B3 (coordinates [8,5]) to exit buffer area D7 (coordinates [31,33]). The dispatch module assigns the task to the transfer device 100 based on historical task weights and device status. After initializing the Q-table, the path planning module begins path training, iterating with a static obstacle count of 6. In the first round, the system uses an exploration strategy of ε = 0.2. After 600 rounds of training, the average path cost decreases from the initial 24.3 steps to 14.8 steps.

[0041] During the execution phase, the transfer device 100 starts and travels to area B3. The tilting arm 121 extends horizontally, the proximal arm 122 descends, and the gripping unit 130 is controlled to complete the gripping. After the sensor determines that the gripping status is successful, it returns a signal, and the scheduling module initiates the second-stage path. The device travels along the predetermined path to the vicinity of area D7, adjusts the attitude of the proximal arm 122, and slowly descends. The gripping unit 130 releases its jaws to complete the placement. The entire process takes 28.4 seconds, with a mission success rate of 98.6% and an error of less than ±12mm, meeting the requirements for industrial-grade transfer accuracy.

[0042] This embodiment demonstrates that the present invention not only possesses high mechanical flexibility and structural stability, but also exhibits excellent learning capabilities and dynamic adaptive abilities in intelligent control and path planning. The system is suitable for intelligent logistics scenarios involving multiple points, batches, and concurrent tasks, and is particularly well-suited for the flexible material flow needs of unmanned warehouses, smart factories, and urban delivery nodes.

[0043] See Figure 2 This is a control flowchart for an intelligent logistics transfer device processing system. In this embodiment, the overall system flow includes multiple stages such as task information access, task allocation logic generation, path planning execution, control command issuance, grabbing / transfer action implementation, and task completion status feedback. The system initially receives order data to be processed through a host computer or cloud platform. This data may include task metadata such as material type, starting location, target location, and time priority. The task allocation submodule matches and evaluates the number of available transfer devices in the system with their respective status indicators (location, power consumption, task load, etc.), and selects the target execution device based on scheduling priority. Subsequently, this module passes the task target and current status as input to the path planning submodule.

[0044] The path planning submodule uses a warehouse scenario represented by a coordinate grid as its modeling basis. Combining obstacle information and task objectives, it employs a Q-learning algorithm for path decision-making. Upon startup, the module initializes the Q-table and uses a greedy strategy with ε=0.3 in the first round to guide a balance between exploration and utilization. Each path training iteration includes state selection, action execution, reward feedback, and Q-value updates, ultimately forming a stable Q-policy matrix within a limited number of rounds. After path planning is complete, control commands are issued to the corresponding conveyor, including the rotation angle, direction of travel, speed limits, and obstacle avoidance mechanisms for each step. Upon receiving the control commands, the conveyor begins operation, sequentially performing movement, grasping, arm adjustment, and deployment operations, and reporting its status back to the scheduling module in real time, forming a closed-loop task management system. This process structure possesses extremely high system stability and multi-machine scheduling flexibility, making it suitable for automated sorting, loading, and transfer operations in large-scale logistics environments.

[0045] See Figure 3 This is a flowchart illustrating the Q-learning algorithm control process within the intelligent collaborative scheduling module of this system. This embodiment employs a model-free reinforcement learning method based on value iteration. The system divides the storage environment into a 40×40 discrete state space, where each state represents a passable or obstructed grid point. The action space includes movement in four directions: up, down, left, and right. Initially, the system assigns a uniform Q-value to all state-action pairs and sets the learning rate α = 0.6, discount factor γ = 0.9, exploration rate ε = 0.3, and a decay factor δ = 0.97 per round. During actual training, the agent (i.e., the transfer device) starts from the initial position, selects the current action according to the ε-greedy policy, and receives an immediate reward after execution. If the action is moved to a non-target point, the reward is -1; if an obstacle is encountered, the reward is -100; and successfully reaching the target position grants a reinforcement reward of +100. After each action, the system updates the Q-table according to the Q-learning formula and updates the current position state s to s', repeating this process until the task objective or the maximum step size limit is reached.

[0046] In a typical experiment, the system simulated 200 tasks, including delivery from the storage area to 10 different unloading points, with an initial average path cost of 23.6 steps. After 400 rounds of Q-value updates, the system converged, ultimately reducing the average path to 12.3 steps. The task success rate increased from 88% to 97%, and the transfer time decreased from an average of 42 seconds to 27 seconds, with path fluctuations less than ±2 grids. This demonstrates that the path planning submodule possesses excellent adaptability and path optimization, dynamically constructing efficient path execution schemes without relying on environmental models, making it particularly suitable for scenarios with multiple obstacles, complex topologies, or dynamic material updates.

[0047] This embodiment further demonstrates that the intelligent logistics transfer device provided by this utility model has a complete structure, closed-loop data and real-time feedback capabilities at the system level, and trainability, optimization and generalization scheduling capabilities at the algorithm level. The full-process control system formed by the two can be widely used in various scenarios such as new intelligent warehousing, intelligent manufacturing plants, and unmanned distribution centers, and has obvious technical advantages in terms of efficiency, stability and intelligence.

Claims

1. An intelligent logistics transfer device, characterized in that, include: The transfer device (100) specifically includes: a transport vehicle unit (110), including a loading section (111) and a column (112) extending upward from the loading section (111); a swing arm unit (120), one end of which is connected to the column (112), the swing arm unit (120) including an inclined arm (121) and a proximal arm (122), the inclined arm (121) and the column (112) being rotatably connected, and the inclined arm (121) and the proximal arm (122) being rotatably connected along the longitudinal axis; a sensor group (113); a gripping unit (130), connected to the distal end of the proximal arm (122); and an intelligent collaborative scheduling module, the intelligent collaborative scheduling module controlling the movement paths of the swing arm unit (120) and the gripping unit (130) according to task instructions.

2. The intelligent logistics transfer device according to claim 1, wherein, The gripping unit (130) specifically includes: unloading action; the unloading action is: the swing arm unit (120) drives the transported object (101) from the location of the transport vehicle unit (110) to the target unloading position, and disconnects the connection between the transported object (101) and the transported object (101) at the target unloading position.

3. The intelligent logistics transfer device according to claim 1, wherein, The gripping unit (130) specifically includes a gripping action; the gripping action is: the gripping unit (130) grips the transfer object (101) at the target location, and the swing arm unit (120) drives it to transfer to the transport vehicle unit (110) to complete the placement operation.

4. The intelligent logistics transfer device of claim 1, wherein, The intelligent collaborative scheduling module includes a task allocation submodule, whose input parameters include: the number of sensor terminal components, the list of tasks to be allocated, the learning rate, the discount factor, the exploration rate and the exploration rate decay coefficient, and dynamically allocates the task objectives of each terminal component.

5. The intelligent logistics transfer device according to claim 4, characterized in that, The intelligent collaborative scheduling module further includes a path planning submodule, which generates the optimal task path based on the Q-learning algorithm and transmits the path instructions to each transfer device.

6. The intelligent logistics transfer device according to claim 5, wherein, The task allocation submodule and the path planning submodule communicate with each other through a data stream and module interface. The task allocation submodule transmits the task allocation results of each sensing terminal component to the path planning submodule.

7. The intelligent logistics transfer device of claim 1, wherein, The intelligent collaborative scheduling module receives external order information as a data source and generates control commands based on the scheduling results to drive the transfer device to complete the task execution; multiple transfer devices form a collaborative cluster, which is controlled by a unified intelligent collaborative scheduling module to achieve multi-task synchronous allocation and path conflict elimination optimization.

Images