Path planning using hybrid grid representation on rough terrain

Path planning approaches based on conventional occupancy grid maps are problematic in off‐road environment because impossible areas include not only obstacles but also landscapes like ramps and pits. The purpose of this paper is to develop a path planning method in a hybrid grid map, which aims to provide a better solution for outdoor navigation.

A hybrid vision system which consists of one stereo vision and one omnidirectional vision is adopted to provide environmental information for 2.5D grid and 2D grid mapping, respectively. An improved planning method originated from conventional D*‐based search algorithm is proposed for more efficient navigation in such hybrid grid maps.

It is confirmed by simulations and experiments that the path planning in the hybrid grid map is more efficient than that in conventional grid maps. Furthermore, it helps to guarantee a safe exploration for field and planetary robots.

This paper proposes a path planning approach in a hybrid grid map representing unstructured environment. The map consists of two different grid representations with diverse resolutions and structures, named 2.5D and 2D grids. The navigation process is expected to become efficient by reducing the replanning times and track length.