CN-122022684-A - Unmanned warehouse cargo allocation autonomous navigation scheduling method and system

CN122022684ACN 122022684 ACN122022684 ACN 122022684ACN-122022684-A

Abstract

The application discloses a warehouse unmanned cargo allocation autonomous navigation scheduling method and a system, wherein the method comprises the following steps of S1, warehouse environment modeling and task analysis; S2, hierarchical collaborative task scheduling, S3, space-time reservation map construction and conflict pre-resolution, S4, dynamic execution and real-time monitoring, S5, multi-objective performance evaluation and self-adaptive parameter scheduling. The application adopts a layered collaborative architecture of upper central dispatching and lower distributed autonomous navigation, the upper layer carries out global task allocation and coarse granularity path planning based on a multi-objective optimization model and a hybrid heuristic algorithm fused with whale optimization and greedy strategy, and the lower layer robot is responsible for local perception, real-time obstacle avoidance and fine control, thereby not only ensuring the overall efficiency of the system to be optimal, but also giving a single machine flexibility to cope with dynamic environment and effectively balancing the requirements of global optimization and real-time response.

Inventors

NIU LIJUAN
MA ZHEXUAN

Assignees

中环低碳节能技术(北京)有限公司

Dates

Publication Date: 20260512
Application Date: 20260129

Claims (10)

1. The unmanned warehouse cargo allocation autonomous navigation scheduling method is characterized by comprising the following steps of: s1, modeling a storage environment and analyzing tasks, constructing a reconfigurable grid storage map, analyzing a logistics task into a triplet sequence of a starting point, an operating point and an end point, and generating a dynamic task queue; s2, hierarchical collaborative task scheduling, wherein an upper central scheduler performs overall task allocation and coarse-granularity path planning based on a multi-objective optimization model and a mixed heuristic algorithm, and a lower layer is deployed on a distributed autonomous navigation unit of each robot to receive instructions and is responsible for local perception, planning and feedback; s3, constructing a space-time reservation map and resolving conflict in advance, generating a space-time track for a planning path, constructing and maintaining a global space-time reservation map to record the time occupation state of each space grid, performing conflict detection in the planning stage, and resolving by adopting a priority-based negotiation adjustment strategy; s4, dynamically executing and monitoring in real time, wherein the robot executes tasks according to reserved paths and reports states in real time, the central system monitors global progress and dynamically responds to delay, faults and newly-added task events, and a rolling optimization window is adopted to periodically reschedule and optimize paths of non-executed tasks; S5, multi-objective performance evaluation and self-adaptive parameter adjustment, wherein multi-dimensional performance indexes are acquired during system operation, and based on a reinforcement learning mechanism, the cost function weight and algorithm internal parameters of the scheduling model are self-adaptively adjusted.
2. The method and system for autonomous navigation scheduling of unmanned warehouse cargo allocation according to claim 1, wherein in S1, the constructing a reconfigurable rasterized warehouse map specifically includes: establishing a global coordinate system based on the warehouse physical layout, and discretizing the warehouse plane area into a uniform grid unit set; Assigning type attributes to each grid unit, wherein the type attributes comprise a goods shelf occupation unit, a two-way passage unit, a one-way passage unit, a warehouse entry unit, a warehouse exit unit, a charging station unit and a static obstacle unit; And storing state information of each grid unit, wherein the state information at least comprises a unit type, coordinates, traffic cost and a real-time occupation mark.
3. The method and system for autonomous navigation scheduling of unmanned warehouse allocation according to claim 1, wherein in S2, the upper-layer central scheduler performs global task allocation by using the following weighted comprehensive cost function: Wherein, the Calculating estimated execution time based on task estimation and completion time of a current task queue of the robot; obtaining the shortest path distance from a starting point to an operating point and from the operating point to an end point, which are calculated or estimated in advance, for the path length cost; estimating the energy consumption cost according to the path distance, the estimated steering times and the acceleration and deceleration model; The cost is estimated for the collision.
4. The unmanned warehouse cargo allocation autonomous navigation scheduling method and system according to claim 3, wherein the hybrid heuristic algorithm merges a whale optimization algorithm and a greedy strategy, and comprises: the method comprises the steps of initializing, namely randomly generating an initial task sequence containing tasks to be distributed for each robot to serve as a position vector of a whale individual; in the whale optimizing stage, simulating the surrounding, bubble net attack and random search behaviors of whales at the head of the whale, updating the individual position, and carrying out global exploration by taking the weighted comprehensive cost as an fitness function; and in the greedy strategy local optimization stage, local fine adjustment is carried out on the optimal individual and the neighborhood individuals thereof in a mode of traversing the task sequence and trying to exchange or insert tasks until the weighted comprehensive cost cannot be reduced.
5. The method and system for autonomous navigation scheduling of unmanned warehouse cargo allocation according to claim 1, wherein in step S3, the resolving using a priority-based negotiation adjustment strategy specifically includes assigning a dynamic priority to a robot task that is in conflict according to a preset rule; if the adjustment fails, rescheduling of the tasks related to the conflict is triggered.
6. The method and system for autonomous navigation scheduling of unmanned warehouse cargo allocation according to claim 1, wherein in S4, the rolling optimization window mechanism specifically comprises: Triggering an optimization flow with a fixed time period or an event period; focusing on all task subsets in a waiting state, and combining the current positions and states of all robots as new initial conditions; and utilizing the mixed heuristic algorithm to redistribute and route plan the unexecuted task subset, and adopting a new scheme superior to the original scheme.
7. The method and system for autonomous navigation scheduling of unmanned warehouse cargo allocation according to claim 1, wherein in S5, the adaptive parameter scheduling mechanism based on reinforcement learning comprises: Modeling a parameter optimization process of a scheduling system as a sequential decision problem, wherein the state is a performance index vector in a statistical period, the action is adjustment of key parameters of a scheduling algorithm, and the reward is a comprehensive feedback value based on multi-objective performance calculation; and outputting new parameter configuration according to the running state of the system by the self-adaptive parameter-adjusting intelligent body, and training by utilizing a deep reinforcement learning algorithm so as to realize the dynamic and personalized optimization of parameters.
8. An unmanned warehouse cargo allocation autonomous navigation scheduling system, comprising: The environment modeling and management module is used for constructing and maintaining a reconfigurable rasterized warehouse map, a shelf information database and a central dynamic task queue; The central cooperative scheduling module is used for executing global task allocation and coarse-granularity path planning based on a multi-objective optimization model and a mixed heuristic algorithm and integrating a conflict pre-detection function; the distributed autonomous navigation module is deployed on each mobile robot, receives global instructions, performs real-time obstacle avoidance and path re-planning based on local perception, and feeds back execution states; The space-time reservation map management module is used for constructing and maintaining a global space-time reservation map and recording the time occupation state of each space grid; the monitoring and self-adaptive optimizing module is used for monitoring the running state of the system in real time, identifying the abnormal execution event and self-adaptively adjusting the parameters of the scheduling model based on the reinforcement learning mechanism.
9. The warehouse unmanned cargo allocation autonomous navigation scheduling system of claim 8, wherein the central co-scheduling module comprises: the task allocation unit runs a mixed heuristic algorithm fused with a whale optimization algorithm and a greedy strategy, and performs robot-task matching with the minimum weighted comprehensive cost as a target; the global path planning unit plans an initial collision-free path for the distributed task and integrates space-time constraint inspection in the planning process; and the conflict pre-detection unit converts the planned path into a space-time track and compares the space-time track with the global space-time reservation map to pre-judge potential conflicts.
10. The method and system for autonomous navigation scheduling of unmanned warehouse cargo allocation according to claim 8, wherein the monitoring and adaptive optimization module is specifically configured to: Receiving state report information of all robots, and carrying out progress tracking, deviation calculation and abnormal event identification and classification; Starting a dynamic response mechanism for task delay, robot faults or newly-added tasks according to the identified event types; and (3) operating the self-adaptive parameter adjusting agent based on reinforcement learning, taking historical and real-time performance data as input, and outputting an adjusting action on the parameters of the scheduling algorithm so as to realize self-optimization of long-term operation performance.

Description

Unmanned warehouse cargo allocation autonomous navigation scheduling method and system Technical Field The application belongs to the technical field of intelligent warehouse logistics automation, and particularly relates to a warehouse unmanned cargo allocation autonomous navigation scheduling method and system. Background With the rapid development of electronic commerce and intelligent manufacturing, warehouse logistics automation has become a key link for improving the efficiency of a supply chain. Traditional warehouse operation mode highly relies on manual operation, faces the challenges of rising of labor cost, unstable operation accuracy, difficulty in continuous improvement of operation efficiency and the like. In recent years, unmanned warehouse systems based on mobile robots gradually become research hotspots and industrialization directions, and automatic carrying and sorting of cargoes are realized through AGVs/AMRs (automatic guided vehicles/autonomous mobile robots) with autonomous navigation, so that the automation level of warehouse operations is remarkably improved. However, the existing unmanned warehouse cargo allocation system still faces a plurality of technical bottlenecks in practical application, namely, in a high-density and dynamically-changed warehouse environment, path conflict and deadlock are easy to occur when multiple robots work cooperatively, the overall efficiency of the system is difficult to ensure by a traditional scheduling strategy based on a fixed path or simple avoidance, and the overall operation efficiency is limited due to the fact that the existing scheduling algorithm adopts a single optimization target (such as a shortest path) and fails to fully consider multidimensional factors such as task timeliness, energy consumption balance, system load balance and the like. In the aspect of navigation technology, the traditional method often adopts a centralized or fully distributed extreme architecture, the former has heavy calculation load and single-point fault risk, the latter has difficulty in ensuring global optimization effect, and meanwhile, the environment modeling mostly adopts a static map and cannot effectively support dynamic obstacle treatment and temporary channel adjustment. At the scheduling algorithm level, the heuristic algorithm (such as a and Dijkstra) has low calculation efficiency in a large-scale multi-robot scene, and the meta-heuristic algorithm (such as genetic algorithm and particle swarm optimization) often has problems of local optimization and slow convergence speed. Therefore, there is a need for a warehouse unmanned cargo allocation navigation scheduling method and system that has global optimization and local adaptability and can be independently learned and adjusted, so as to cope with complicated and changeable modern logistics warehouse requirements. Disclosure of Invention The application provides a warehouse unmanned cargo allocation autonomous navigation scheduling method and system, and aims to solve the problems of low efficiency and limited overall operation efficiency in the prior art. In a first aspect, a warehouse unmanned cargo allocation autonomous navigation scheduling method includes: s1, modeling a storage environment and analyzing tasks, constructing a reconfigurable grid storage map, analyzing a logistics task into a triplet sequence of a starting point, an operating point and an end point, and generating a dynamic task queue; s2, hierarchical collaborative task scheduling, wherein an upper central scheduler performs overall task allocation and coarse-granularity path planning based on a multi-objective optimization model and a mixed heuristic algorithm, and a lower layer is deployed on a distributed autonomous navigation unit of each robot to receive instructions and is responsible for local perception, planning and feedback; s3, constructing a space-time reservation map and resolving conflict in advance, generating a space-time track for a planning path, constructing and maintaining a global space-time reservation map to record the time occupation state of each space grid, performing conflict detection in the planning stage, and resolving by adopting a priority-based negotiation adjustment strategy; s4, dynamically executing and monitoring in real time, wherein the robot executes tasks according to reserved paths and reports states in real time, the central system monitors global progress and dynamically responds to delay, faults and newly-added task events, and a rolling optimization window is adopted to periodically reschedule and optimize paths of non-executed tasks; S5, multi-objective performance evaluation and self-adaptive parameter adjustment, wherein multi-dimensional performance indexes are acquired during system operation, and based on a reinforcement learning mechanism, the cost function weight and algorithm internal parameters of the scheduling model are self-adaptively adjusted. Optionally, in the st