Moving Obstacle Detection And Avoidance For Unmanned Ground Vehicle

Abstract

This thesis presents the development and implementation of an autonomous obstacle avoidance algorithm for UGV (Unmanned Ground Vehicle). The research improves the prior work by enhancing the obstacle avoidance capability to handle moving obstacle as well as stationary obstacles. A mathematical representation of the area of operation with obstacles is formulated by PTEM (Probabilistic Threat Exposure Map). The PTEM quantifies the risk in being at a position in an area with different types of threat. Threat in this context means UGV getting close to or running into stationary and moving obstacles. A LRF (Laser Range Finder) sensor mounted on the UGV is used to collect information about the obstacles in the area. LRF readings are used to construct the PTEM. A guidance algorithm processes the PTEM and generates the guidance (speed and heading) commands to steer the UGV to assigned waypoints while avoiding obstacles. The main contribution of this research work is to update the PTEM continuously as new LRF reading are obtained. With this approach, no change is needed in the guidance algorithm since the PTEM will have representation of the obstacles at any given time and the guidance algorithm processes the updated PTEM. The improved PTEM construction algorithm is implemented in a MATLAB/Simulink simulation environment that includes models of the UGV, LRF, all the sensors and actuators needed for the control of the UGV. The performance of the algorithm is also demonstrated in realtime experiments with an actual UGV system.