Integration of 3D Lines and Points in 6DoF Visual SLAM by Uncertain Projective Geometry

In this paper we face the issue of fusing 3D data from different sensors in a seamless way, using the unifying framework of uncertain projective geometry. Within this framework it is possible to describe, combine, and estimate various types of geometric elements (2D and 3D points, 2D and 3D lines, and 3D planes) taking their uncertainty into account. By means of uncertain projective geometry, it is possible to derive simple bilinear expressions to represent join and intersection operators using only three matrices as operator. In particular, we are interested in using 3D information coming from different (logical) vision sensors observing the same scene, to improve map accuracy. The experimental section shows that it is possible to improve both mapping accuracy and pose estimation while performing SLAM with a mobile robot, by integrating sensor information coming from trinocular feature-based vision and correlation based stereo.