Autonomous following of ill-defined roads is an important part of visual navigation systems. The majority of current image-based road following methods rely on computationally expensive algorithms. This paper presents an adaptive real-time method based on statistical analysis of the colour of a road surface in a trapezoidal shape that corresponds to the projection of the road on the image plane. Our method is capable of navigating in real-time in a variety of situations, including 90°turns and crossroads, and coping with variable conditions of the road such as surface type and shadows.
|Title of host publication||Advances in Autonomous Robotics|
|Subtitle of host publication||Joint Proceedings of the 13th Annual TAROS Conference and the 15th Annual FIRA RoboWorld Congress, Bristol, UK, August 20-23, 2012|
|Publication status||Published - 2012|