Ground plane detection for autonomous vehicle in rainforest terrain

This paper presents the determination of the ground plane method for navigation in rainforest terrain using stereo vision. Ground plane detection plays an important role for obstacle detection task as it act as the reference plane of the scene. V-disparity image is widely used to detect the ground plane for on-road and off-road navigation. However, it relies on distinct road features such as lines and edges which may not exist in rainforest terrain. In this work, we attempted to use colour clustering method to extract possible ground profile even without using distinct road features.