Formation Control of Time-varying Multi-agent System Based on BP Neural Network

Abstract

In this paper, the formation control problem of time-varying multi-agent system (MAS) based on back propagation neural network (BPNN) is investigated. The system consists of two robots, which effectively improve the system operation efficiency through the coordination of robots. Moreover, leader-follower based strategy is used to formation. Among them, the leader robot can be controlled wirelessly through blue-tooth to share monitoring data in real time. The front image can be captured by the on-board camera mounted on the follower robot and optimized by the least square method (LSM). Then machine vision is used to locate the leader robot and the follower robot's position is adjusted in real time through proportion integral differential (PID) cnotroller based on BPNN to realize the formation. Compared with the traditional PID controller, BPNN PID cnotroller takes the weight coefficient of NN as an adaptive parameter, only few parameters need to be updated online, which can greatly lessen the computational burden. Finally, simulation experiments are given to illustrate the results.