Optimal Local Path Planner Over Receding Horizon Using Open Interval B-Spline

Abstract

A local path planning algorithm aims to provide a global, deterministic and safe solution for the dynamic navigation of wheeled robots with limited visibility. To this end, a novel approach based on open interval B-spline curves computed over a receding horizon is exposed in this paper. Real-time performances are ensured by using an interval branch and bound algorithm. The resulting path is smooth, obstacle-avoidant, and continuously connects local paths without requiring additional computation. Furthermore, this approach offers a guaranteed understanding of the solution's state. A large set of simulations, adapted for a wheeled differential robot on several scenarios, is finally carried out to assess parameters impact and performances.