Lane following system for a mobile robot using information from vision and odometry

The primary objective of our lane following system is to autonomously drive a mobile robot between two parallel lanes. Usually, the low sampling rate and the visibility of the environment are the main limitations of the vision sensors in a lane following system. In this paper, a higher sampling rate odometry sensor is integrated with a vision-based system using an Unscented Kalman Filter (UKF). The lane following system is implemented on a Pioneer 2 mobile robot and the experimental results indicate the applicability of the approach.