Road trajectory mining and autonomous steering control for vision-based unmanned vehicles

Autonomous steering control is the principal task in the development of an intelligent transportation system. This research paper proposes a novel approach for vision based intelligent control of unmanned vehicles. The paper addresses the problems of accurate and efficient intelligent vehicle control by incorporating a well known evolutionary algorithm cAnt-Miner. The uniqueness of the proposed algorithm lies in its ability to accurately steer the vehicle on structured and unstructured roads in real-time, without requiring meticulous calculation. The algorithm first uses the road edges to evaluate the direction of the path through classification via cAnt-Miner. Subsequently, it steers the vehicle according to the particular direction using fuzzy logic controller. The synergy of the evolutionary algorithm, cAnt-Miner, for trajectory mining with fuzzy logic control enables autonomous steering with high efficacy. Experiments were carried out on our university´s Intelligent DRIving System (IDRIS). Experimental results show that the proposed algorithm can steer the car automatically in real-time.