Free-space Polygon Creation based on Occupancy Grid Maps for Trajectory Optimization Methods

Abstract

Abstract In this paper, we present a strategy for incorporating occupancy grid maps into local trajectory optimization methods to allow for safe autonomous navigation. We construct polygons as surface representations of the locally visible area and merge them with the aid of third-party software to form connected free-spaces. The method can be applied to any two-dimensional environments and different representations of vehicle dimensions. It allows direct optimization methods to be used consistently for different problem levels from control to planning. The strategies developed are applied to a rover system that moves in the plane. We evaluate our algorithms on the basis of simulation results and real-world experiments.