Robot localization using a complementary laser/camera filter

In this paper, we propose a novel localization method for indoor-wheeled robots. The system consists of fusing scene and range data to make more robust the 3D-to-3D egomotion estimation, which is typically done via ICP. To validate our approach and assess its performance, a system comprised of a laser range finder paired with a monocular camera is implemented and several experiments are performed. Results show that, provided the ground is flat, our proposed system is applicable to unstructured indoor environments while previous hybrid techniques are not.