A method for border patrolling navigation of a mobile robot

In this paper, we propose a control law for patrolling the edge of an obstacle, that considers sampled line-of-sight obstacle distance data. By forming some assumptions about the shape of the obstacle, we generate constraints suitable for navigation using a model predictive control type approach. We also show how a target point may be generated to enable navigation in-line with our goal of border patrolling. The proposed method is suitable for vehicles with unicycle dynamics, and has the advantage of being able to vary the vehicles speed and patrolling distance to adapt to the obstacle. We are able to show collision avoidance, complete transversal of the obstacle and finite completion time for transversing a finite boundary segment. Simulations confirm the validity of the method.