Perception For Rugged Terrain

To perform planetary exploration without human supervision, a complete autonomous robot must be able to model its environment and locate itself while exploring its surroundings. To this end, estimating motion between sensor views and merging the views into a coherent map are two important problems. In this paper, we present a 3-D perception system for building a geometrical description of rugged terrain environments from range data. We propose an intermediate representation consisting of an elevation map that includes an explicit representation of uncertainty and labeling of the occluded regions. We present an algorithm, called the Locus method, to convert range image to an elevation map. An uncertainty model based on this algorithm is developed. Based on this elevation map and the Locus method, we propose a terrain matching algorithm which does not assume any correspondences between range images. The algorithm consists of two stages: First, a feature-based matching algorithm computes an initial transform. Second, an iconic terrain matching algorithm that minimizes the correlation between two range images is applied to merge multiple range images into a uniform representation. We present terrain modeling results on real range images of rugged terrain.