In this paper, we describe a system that can carry4 out simultaneous localization and mapping (SLAM) in large in-5 door and outdoor environments using a stereo pair moving with 66 DOF as the only sensor. Unlike current visual SLAM systems that7 use either bearing-only monocular information or 3-D stereo in-8 formation, our system accommodates both monocular and stereo.9 Textured point features are extracted from the images and stored10 as 3-D points if seen in both images with sufficient disparity, or11 stored as inverse depth points otherwise. This allows the system to12 map both near and far features: the first provide distance and ori-13 entation, and the second provide orientation information. Unlike14 other vision-only SLAM systems, stereo does not suffer from "scale15 drift" because of unobservability problems, and thus, no other in-16 formation such as gyroscopes or accelerometers is required in our17 system. Our SLAM algorithm generates sequences of condition-18 ally ...
Lina María Paz, Pedro Pinies, Juan D. Tard&