Data-driven time-varying formation-containment control for a heterogeneous air-ground vehicle team subject to active leaders and switching topologies

Abstract

The optimal formation-containment control problem for a team of heterogeneous unmanned air-ground vehicles (UA-GVs), subject to active leaders and switching topologies, is addressed via reinforcement learning. The quadrotors are introduced to achieve predetermined time-varying formation and the ground vehicles are designed to move into the convex hull spanned by the quadrotor formation. The quadrotor dynamics is underactuated, and the UA-GV system involves nonlinear dynamics and uncertain dynamical parameters. Distributed observers are developed for each vehicle to provide the position reference under the effects of switching topologies and unpredictable maneuvers of the leaders. Optimal control laws are proposed without accurate information of the dynamical models of the UA-GVs using reinforcement learning. Simulation results of a heterogeneous UA-GV team are presented and the superiority of the proposed data-driven optimal control laws is validated.