Distributed Rendezvous And Tracking For Multiple Unicycles With Heterogeneous Input Disturbances

A multi-agent system is a group of static or mobile agents deployed in a geograph-ic area to perform collective tasks.It has drawn much attention in various fields recently due to its superior performance and a wide range of application scenarios.One problem that plays a central role in the study of multi-agent system is the consensus problem,which already has many in-depth studies,but there are still some limitations.Firstly,distributed control of nonholonomic systems has been less studied due to the com-plexity of nonholonomic constraints.Secondly,most studies on disturbance observer based control focus on linear systems;the topic has already been investigated for non-linear systems.Finally,most works on cooperative control do not consider collision avoidance;the research results more focus on the theoretical perspective,without the application or experiment in practical robots.This paper addresses the distributed rendezvous and tracking problems for mul-tiple unicycles in the presence of unknown heterogeneous disturbances.Disturbance compensators are designed to attenuate the effect of unknown disturbances,which rely on a state estimator constructed to estimate the states of the controlled agents.LaSalle's invariance principle and Barbalat's lemma are employed,respectively,to prove the con-vergence of the systems.For the rendezvous problem,all agents will rendezvous at a pre-specified point eventually under the effect of input disturbances.For the tracking problem,all follower agents will track the leader agent and then move with the leader at the same linear and angular velocities under the effect of input disturbances.Then,the potential function approach is employed in constructing the control laws that are capable of avoiding inter-agent collisions.With this method,the tracking error can be made upper bounded while all the agents keep a "safe" distance from each other.Simulation examples are then presented to verify the effectiveness of the designed con-trollers.Meanwhile,a software based on MFC is developed to apply the control laws on AmigoBots,which also demonstrates the validity of the theoretical results.