Multi-agent motion planning and coordination in polygonal environments using vector fields and model predictive control
Rashmi Hegde, Dimitra Panagou · 2016
In this paper, we extend earlier work on motion planning and coordination of multiple agents, to environments of arbitrary polygonal obstacles, using non-gradient vector fields to steer each agent towards their goal configurations while avoiding collisions. We formulate the vector fields so that the pattern of their integral curves depends on a parameter λ. By manipulating the value of λ, we obtain a set of vector fields whose integral curves define the flow lines for an a priori known obstacle environment. We use the vector field design in tandem with model predictive control to compute safe trajectories for multi-agent systems. The competence of the proposed methodology is demonstrated for both static and dynamic environment via simulation results. The efficacy of model predictive control in achieving control trajectories, free of chattering, for multi-agent coordination is validated through comparison to a state feedback coordination and control protocol.