Research On Task Assignment And Path Planning Of Multi-robots System In Smart Factory

Multi-Robots System(MRS)has good autonomous mobility,sensitivity,and strong robustness,overcoming the bottlenecks faced by a single Robot operation.It is widely used in various fields of industrial production.Multi-robots system how to complete the smart factory high real-time data patrol tasks have always been the hot spot of multi-robots system research.In this thesis,the cooperative multi-robots system for data patrol and studies the path planning problem of task allocation,reasonable task allocation,and path planning are not only reflected the significance of existence in the multi-robots system it also meets the high real-time requirements of intelligent operation in the era of “Industry 4.0”.When multi-robots systems cooperate to perform data patrol,the primary problem to be solved to achieve the completion of global tasks on time is to allocate tasks reasonably for each robot and realize the load balance of multi-robots systems.When the multi-robots system obtains the optimal task assignment strategy,it will perform the patrol task in the corresponding workshop according to the assignment strategy.At this time,there may be congestion,collision,and conflict phenomenon in the working environment,so how to plan a conflict-free optimal path for each robot in the complex and changeable environment is particularly important.The research contents of this thesis are as follows:1)Aiming at the problem that the existing task assignment algorithm failed to consider the mapping relationship between the resource status of multi-robots and the resource demand of multi-patrol tasks or failed to adapt to the dynamic environment,this thesis proposed a task assignment strategy of multi-robots system based on Stackelberg dynamic game.The dynamic behavior of the participants in the Stackelberg game was exactly in line with the dynamic interaction scenario of multi-robots and multi-patrol tasks.Firstly,the utility function of load balancing was set for multi-robots system based on the game model,and the maximum utility function was taken as the goal of the game to realize the load balancing of multi-robots system.Then,an improved reinforcement learning SARSA algorithm was proposed to continuously learn the optimal task assignment strategy to meet the high real-time demand of the smart factory patrol tasks.In order to adapt to the dynamic change of task assignment strategy,state transition probability and action transition probability were introduced,and the deterministic strategy of SARSA algorithm was transformed into random strategy.In order to obtain the best task assignment scheme of each robot as soon as possible,the average state action function was introduced to monitor the state action value,and the learning rate of SARSA algorithm was adjusted dynamically.2)Aiming at the complex working environment of multi-robots system and the problems such as large amount of computation and poor real-time performance of path planning algorithm,a global path planning algorithm without conflict based on dynamic weight matrix was proposed to plan the optimal path without conflict time for multi-robots connecting multiple nodes.Firstly,in order to solve the problem of path congestion or collision caused by the increasing number of robots,a central controller was introduced to observe the positions of other robots in real time,and the path weight matrix was dynamically updated to avoid the existing path congestion,collision and conflict in advance.And then put forward a kind of time to optimize the global path planning algorithm,the connection was more than the task of inspection for single path in the iterative search contains two nodes weight minimum stretch,then look for other tasks include the sections on both ends of the node and the node time optimization path,and planning a contains the best iterative search for continually checking task set the optimum path of the node of the time.Finally,this thesis completed the multi-robots data patrol tasks based on the multi-robots task assignment and conflict-free path planning strategies.Experimental results showed that the task assignment strategy based on the Stackelberg game achieves utility maximization and environmental adaptation optimization.The time-optimized global path planning algorithm based on a dynamic weight matrix ensured the path conflict-free and realized the optimization of the path movement time of multi-robots,which meet the real-time requirements of industrial tasks and had certain reliability.