Predictive terrain contour mapping for a legged robot

Abstract

Most legged robots have to negotiate unknown environments with little or no descriptive terrain data as autonomous terrain mapping facilities for legged robots are limited. A predictive terrain contour mapping strategy is proposed which considers the use of feed-forward neural networks to predict terrain contours in unstructured environments based on sample data extracted from the walking surface during the locomotion of Robug III - an eight legged, pneumatically powered walking and climbing robot. In simulation, it is shown that the prediction performance is very acceptable; practical tests are conducted on a prototype robot leg and the results are compared with those obtained in simulation.