Randomized continuous monitoring of a target by agents with turn radius constraints

Abstract

We consider a scenario where a target needs to be monitored continuously by a set of agents. The agents are equipped with sensors fixed to their bodies and the sensors have a finite footprint. The mission requires the target to lie in the sensor footprint of at least one agent at any time. Due to the finite footprint, an agent can monitor the target only when its range and orientation with respect to the target is within certain limits. The agents considered here have kinematic motion constraints and limits on their speed. This constraint mandates the use of multiple agents to monitor the target sequentially in order to achieve continuous monitoring. To reduce the predictability of their paths, the agents randomly generate successive monitoring positions. We present strategies for scheduling and path planning which assure continuous monitoring; the path planning is done using Bézier curves subject to curvature constraints, boundary constraints and length constraints. We also present an efficient computation methodology to enforce maximum curvature and path length constraints using only the control points of the curves.