Obstacle & lane detection and local path planning for IGVC robotic vehicles using stereo vision

Abstract

A robotic car, also known as a driverless or self-driving car, is an autonomous vehicle capable of sensing its environment and navigating itself without human input. Robotic cars exist mainly as prototypes and demonstration systems, but are likely to become more widespread in the near future. Usually, autonomous vehicles sense their surroundings forlane detection and obstacle avoidance with radar, 2D LIDAR, 3D LIDAR, and camera sensors, This article describes our approach in developing an affordable stereo vision system using two ordinary webcams and OpenCV (Open Source Computer Vision) library. We applied our stereo vision system to Vulture 2 and iWheels robotic platforms to enter IGVC (Intelligent Ground Vehicle Competition) 2012 and 2013. The results show that stereo vision based navigation is a promising and affordable mechanism to develop robot cars.