Multi-robot Formation Control Based On State Estimation Algorithm

Formation control is an important topic in the multi-robot control domain and has a very wide range of applications. Multi-robot formation system can often complete some tasks effectively at low cost which may be very difficult or even impossible for a single robot. So it is highly significant to study the multi-robot formation control technology.Based on the leader-following approach, this thesis designs a multi-robot formation system by using three Qbots and then applys the consensus algorithm with estimation to the system, so that the formation system can have less deviation in the environment with occasional communication failure. Firstly, aiming at the problem that an obvious tracking deviation may be caused when occasional communication failure appears in a multi-agent tracking system, this thesis presents a consensus algorithm with estimation. The kernal idea of the algorithm is that the follower not only compute the corresponding control with the received reference status but also to estimate the track of the reference status when the communication works well, and the algorithm can estimate the reference status in order to compute the corresponding control when communication failure occurs. This thesis proves its rationality mathematically and also shows its effectiveness by Simulink. Then based on the Qbot flatform of Quanser Company, this thesis models the Qbot and designs a formation system with three Qbots which can let two Qbots rotate around the third one while maintaining an equilateral triangle formation. Finally, this thesis applys the consensus tracking algorithm with estimation to the formation system and then conducts the experiment with artificial communication failure, and then analyses the results.The result shows that the consensus algorithm with estimation can reduce the deviation at certain extent for the formation system with occasional communication failure. Therefore, the effectiveness and the feasibility of the formation algorithm with estimation can also be proved.