Ground Robot Terrain Mapping and Energy Prediction in Environments with 3-D Topography

Abstract

Energy usage for robots operating in off-road environments depends on the type of terrain and its 3-dimensional topography. Unfortunately, the terrain-dependent rolling resistance is typically unknown. This paper presents a method for predicting the energy cost of paths through such environments based on data previously collected by the robot. Using a longitudinal vehicle model and known environment topography, the rolling resistance at a point in the environment is estimated with a Recursive Least Squares (RLS) filter. These point estimates are then used to build a spatial rolling resistance map of the environment with Gaussian process regression (GPR). The map allows for the probabilistic prediction of the energy cost of future missions. Simulation results are provided for the energy prediction and preliminary experimental results on a ground robot platform validate the rolling resistance mapping method.