Multi-robot Task Assignment And Path Planning In Dynamic Environment

Compared with single robot systems,multi-robot systems have better robustness,economic,and scalability.At present,multi-machine systems are mainly used in smart manufacturing,patrol monitoring,search and rescue,and transportation logistics.There are many problems that need to be studied in the multi-robot system.Task allocation and path planning are hot and difficult issues in this field.This dissertation focuses on the research of multi-robot task assignment and routing in dynamic environment.In the actual multi-robot task assignment and path stiffness planning problems,the working environment of multi-robots is often dynamic,for example,the mission position has moved,the obstacle position has changed,and the robot has experienced a sudden failure.Traditional task assignment and path planning algorithms are often incapable of solving such problems or even real-time performance cannot be guaranteed.SOM-based task assignment and path planning algorithms benefit from the good self-organizing,self-learning and self-adaptiveness of the SOM network,which can well solve the problem of task allocation and path planning for multi-robots in a dynamic environment.The algorithm can plan a feasible path for each robot while assigning tasks.The traditional SOM task assignment algorithm has output neuron oscillation phenomenon.This phenomenon will lead to too many iterations of the algorithm and even cause the algorithm to fall into endless loop.Not only that,but it also makes the path planned by th e algorithm become full of turning points.For the output neuron oscillation,this thesis proposes a locking mechanism to avoid this phenomenon.With the two conditions of the winner's no-neighbor condition and the minimum cost of the locking mechanism,th e robot can dynamically lock a certain task.This lock has a certain convergence,each robot in the system will eventually lock a robot to avoid output neuron oscillations.Experimental results show that the locking mechanism can significantly reduce the n umber of iterations of the algorithm under the premise of improving the allocation cost,and avoid the output of the neuron oscillation phenomenon.In order to solve the problem of task allocation and path planning for multi-robots in an obstacle environment,SOM task assignment algorithms,locking mechanisms,and artificial potential fields are integrated in this thesis.On the one hand,the locking mechanism can guarantee that the path planned for each robot is very smooth,and on the other hand,the artificial potential field can ensure that the robot can avoid obstacles flexibly during the course of its progress.In traditional algorithms,task allocation and path planning are often divided into two steps,which means that when some tasks,robots,or obstacles in the environment change,task allocation and path planning need to be re-executed,which leads to the algorithm's Real-time is not very good.In the algorithm proposed in this thesis,the task allocation and path planning are synchronized with the robot,and the algorithm can respond to the environment and the changes of the robot in real time in real time.In order to prove the validity and reliability of the algorithm proposed in this thesis,rich experiments were conducted.The experimental results show that the proposed algorithm can plan a smooth and collision-free safe path for each robot while allocating tasks for each robot,whether it is in a static environment or a dynamic environment.