VISION BASED SAFE ROAD TRACKING SYSTEM FOR AUTONOMOUS GROUND VEHICLE
T. Abd El-Zaher, A. Bayoumy, A.Μ. Sharaf, Y. Hossam El-Din · The International Conference on Applied Mechanics and Mechanical Engineering · 2012
The past decade has witnessed a rapid evolution of the sensors technology as wellas the embedded systems which in turn stimulates the development of a futuristicfield such as the autonomous mobile systems. However, the development of anautonomous ground vehicle that is able to sense and avoid obstacles on its wayrepresents a major challenge up till now. Various autonomous ground vehicles havebeen recently introduced which rely on wide diversity of detection systems such as;laser systems, acoustic systems and machine vision systems.This paper proposes an advanced algorithm for an autonomous ground vehiclebased on lane detection and tracking in addition to obstacle avoidance technique.Through this algorithm, the suitable car heading angle is calculated in order to followthe lane or avoid predefine obstacles. The lane and / or obstacles are beingrecognized based on vision system in which a real time data are collected by asingle camera and analyzed using the color separation technique. An in-housenumerical model is developed to estimate both the relative distance and direction ofthe obstacle. Consequently, further actions are considered to enable the vehicle todrive accurately down a predefined path and at a defined speed. The results revealboth the fidelity and the robustness of the introduced algorithm.