An internal model approach to autonomous leader/follower trailing for non-holonomic vehicles

The focus of this paper is on the design of a control architecture of decentralized type for controlling a leader/follower pair of autonomous non-holonomic vehicles. A fundamental constraint in this trailing control requires that each agent employs local sensor information to process data on the relative position and velocity between its neighbouring vehicles, without relying on global communication with mission control. This constraint poses a challenge in the design of the control system because the reference trajectory to be tracked, which in the case considered in this paper is related to the motion of the leader, is not known a priori. It is shown in the paper that this specific control problem can be approached from the point of view of the internal model paradigm. In particular, once models of the autonomous dynamics of the leader are embedded in a decentralized dynamic controller, the design of the controller can be completed with a robust stabilizer, obtained by using ISS-gain-assignment techniques. It is shown that asymptotic convergence of the follower to an arbitrarily small neighbourhood of the desired steady-state configuration is achieved, despite the presence of possibly large parameter uncertainties, while the motion of each agent remains confined into specified 'sectors' to avoid possible collision between neighbouring vehicles during transients. Simulation results are presented to illustrate the design methodology. Copyright (C) 2006 John Wiley & Sons, Ltd.