A feedback-based multiagent navigation framework

In this paper, an on-line multirobot navigation methodology is presented, extending the concept of navigation functions from the single-point-sized holonomic robot domain to the domain of multiple non-point-sized holonomic robots. An appropriate measure of the distance from bad sets, suitable for multirobot navigation, is introduced. The derived closed form analytic solution provides a feedback-based navigation scheme, suitable for implementation, with theoretically guaranteed global convergence and collision avoidance properties. The proposed methodology is validated through non-trivial simulations.