A novel method of sensing and classifying terrain for autonomous unmanned ground vehicles

Prior, Stephen (2015) A novel method of sensing and classifying terrain for autonomous unmanned ground vehicles. Middlesex University, Doctoral Thesis, 141pp.

Abstract

Unmanned Ground Vehicles (UGVs) play a vital role in preserving human life during hostile military operations and extend our reach by exploring extraterrestrial worlds during space missions. These systems generally have to operate in unstructured environments which contain dynamic variables and unpredictable obstacles, making the seemingly simple task of traversing from A-B extremely difficult. Terrain is one of the biggest obstacles within these environments as it could potentially cause a vehicle to become stuck and render it useless, therefore autonomous systems must possess the ability to directly sense terrain conditions. Current autonomous vehicles use look-ahead vision systems and passive laser scanners to navigate a safe path around obstacles; however these methods lack detail when considering terrain as they make predictions using estimations of the terrain’s appearance alone. This study establishes a more accurate method of measuring, classifying and monitoring terrain in real-time. A novel instrument for measuring direct terrain features at the wheel-terrain contact interface is presented in the form of the Force Sensing Wheel (FSW). Additionally a classification method using unique parameters of the wheel-terrain interaction is used to identify and monitor terrain conditions in real-time. The combination of both the FSW and real-time classification method facilitates better traversal decisions, creating a more Terrain Capable system.

This record has no associated files available for download.

Contributors

Author: Stephen Prior

Thesis advisor: Stephen Prior

Download statistics