Terrain prediction for an eight‐legged robot

Abstract

Abstract Most legged robots must negotiate unknown environments with little or no terrain knowledge, as autonomous terrain mapping for robots is limited. A predictive terrain contour mapping strategy is proposed, which employs the use of a feed‐forward neural network to predict the contours in environments, based on the positions of the neighboring legs. The predicted performance is better than previous implementations. © 2002 John Wiley & Sons, Inc.