| With the rapid development of e-commerce and the change of people’s shopping methods,intelligent warehousing system has become an important research and development direction of e-commerce enterprises.The intelligent warehousing system combines digital factories and artificial intelligence,and adopts the picking mode of "goods to people",which greatly improves the production efficiency of the entire factory.Multi-robot system is an important part of intelligent warehousing system,and task assignment is a very critical combinatorial optimization problem in multi-robot system.The result of task assignment is directly related to the efficiency and cost of the warehouse system to complete the order,so the problem of multi-robot task assignment is a research topic with application value.At present,the complexity of the warehousing system continues to increase,and the traditional task assignment method cannot meet the needs of real-time.This paper takes the intelligent warehousing system as the background,and conducts in-depth research on the task assignment problem of multi-robot system.The main research contents include:Aiming at the problem of multi-robot task assignment in static environment,a static mathematical model is established and an extended consensus-based bundle algorithm based on layering is proposed.First,the algorithm designs a new bidding function considering complex constraints such as robot speed,task time window constraints,arrival time and travel cost loss.At the same time,based on the idea of the k-means algorithm in the clustering algorithm,the problem of task assignment is handled in a hierarchical manner,that is,the robots are first assigned to each task area,and then the tasks are assigned to the robot in each task area.Moreover,the proposed algorithm can solve the coexistence of simple tasks and complex tasks that need the cooperation of multiple robots,and there is a certain priority relationship between different tasks.The simulation results show that the algorithm can effectively achieve multi-robot task assignment in a static environment,and can effectively reduce the task assignment time on the basis of ensuring the benefits of task assignment.Aiming at the problem of multi-robot task assignment in dynamic environment,a dynamic mathematical model is established,and a method combining local replanning method based on consensus-based bundle algorithm and improved wolf pack algorithm is proposed.In response to the increase in the number of tasks,the local replanning method based on the consensus-based bundle algorithm is adopted to deal with the situation,which can solve the situation of multiple new tasks of different types and all tasks have a certain priority.In view of the reduction of the number of robots,the improved consensus-based bundle algorithm is difficult to solve,and the improved wolf pack algorithm is used to deal with it.The algorithm incorporates optimization strategies such as scheme selection optimization method,integer matrix coding,genetic algorithm idea,disturbance near the head wolf,reinitialization of the stagnant state of the head wolf and so on.The simulation results show that the algorithm can effectively solve the problem of multi-robot task assignment in a dynamic environment,and its performance has certain advantages compared with other algorithms.In order to further verify the feasibility of the algorithm in a near-real environment,a multi-robot task assignment simulation platform was built to test the performance of the algorithm.The simulation platform consists of Python target controller,Unreal Engine 4,and Microsoft’s open source Air Sim.As the upper-level control system,the Python target controller is associated with the algorithm,and displays the synchronization of robot and task model in real time in the simulation environment.Unreal Engine 4 and Microsoft’s open source Air Sim are used as the underlying physics engine to intuitively display multi-robot motion and task distribution.The algorithm is tested on the simulation platform,and the performance of the algorithm is evaluated by the operation of the robot and the task in the interface.The experimental results show that the proposed algorithm can complete robot clusters to assign tasks in real scenarios. |
