Multi-robot Scheduling Research For Recharging And Motion Uncertainty In Warehouses

With the gradual development and maturity of e-commerce,warehousing and logistics have become an important part worthy of people's attention.In order to improve the transportation efficiency and save manpower and material resources,the use of multi robots systems is widely used in warehouse scenarios to perform tasks such as cargo transportation.However,the uncertainty among multiple robots affects the long-term stable operation of warehousing system.Therefore,the multi-robot scheduling research based on uncertainty is very important to improve the operational efficiency of the storage system and ensure the real-time performance.This paper mainly researches the multirobot scheduling algorithm based on recharging and motion uncertainty.And above research topic is composed of autonomous recharging problem based on recharging uncertainty and multi-robot path planning problem based on motion uncertainty.In this paper,following algorithms were designed to solve above two problems:In order to solve autonomous recharging problem based on recharging uncertainty,this paper designed a multi-robot recharging planning algorithm.This algorithm includes two phases: recharging task planning of individual robot and multi-robot recharging tasks coordination.In the first phase,recharging task planning based on task sequence segmentation is proposed to split the individual robot task sequence into sections and plan the recharging tasks one by one.Then in the second phase,all robots' task sequences with planned recharging tasks are centralized and selected recharging stations of some planned recharging tasks are reallocated,using maximum bipartite graph matching to reduce waiting time for other robots recharging and improve overall task execution efficiency.In this paper,the designed algorithm is theoretically analyzed from the total task completion time and the computation time.In order to solve multi-robot path planning problem based on motion uncertainty,this paper designed a bi-level path planning algorithm.In this paper,warehousing map is partitioned using a discrete modeling method and architecture of proposed algorithm is composed of topology level and route level generating from above map: in the topology level,the order of passing districts is planned combined with the district crowdedness to achieve the district equilibrium and reduce the influence of robots under motion uncertainty.And in the route level,a MDP method combined with probability of motion uncertainty is proposed to plan path for all robots in each district separately.At the same time,the number of steps for each planning is dependent on the probability to decrease the number of planning.The conflict-avoidance is proved and optimization is discussed for the proposed algorithm.In order to verify the effect of the above two algorithms,this paper developed a multi-robot scheduling and simulation system based on the real warehousing environment separately.And the scheduling system have many function modules including: map management module,task management module,AGV management module,The functions of simulation system include: warehouse simulation and AGV simulation.The above two systems take the test of multiple robots operation in this paper,which proves that they can meet greatly meets the makespan and real-time requirements of the warehouse environment.At the same time,,the two algorithms designed in this paper are simulated based on above two systems,and the simulation results verify the effectiveness and real-time performance of the algorithm.