Vision-based Terrain Classification and Classifier Fusion for Planetary Exploration Rovers

Autonomous rover operation plays a key role in planetary exploration missions. Rover systems require more and more autonomous capabilities to improve efficiency and robustness. Rover mobility is one of the critical components that can directly affect mission success. Knowledge of the physical properties of the terrain surrounding a planetary exploration rover can be used to allow a rover system to fully exploit its mobility capabilities. Here a study of multi-sensor terrain classification for planetary rovers in Mars and Mars-like environments is presented. Supervised classification algorithms for color, texture, and range features are presented based on mixture of Gaussians modeling. Two techniques for merging the results of these "low level" classifiers are presented that rely on Bayesian fusion and meta-classifier fusion. The performances of these algorithms are studied using images from NASA's Mars Exploration Rover mission and through experiments on a four-wheeled test-bed rover operating in Mars-analog terrain. It is shown that accurate terrain classification can be achieved via classifier fusion from visual features. Thesis Supervisor: Karl lagnemma Title: Principal Research Scientist