At the 2002 ICRA meeting, I presented a paper on how computer vision techniques can be applied to the bearings-only Simultaneous Mapping and Localization (SLAM) problem, in order to obtain a linear algorithm that recovers both robot poses and
observed landmarks. Ashley Stroupe did a series of experiments with the Winnow Robots in order
to validate the method experimentally, which is also described in the paper.
The method supplies a good initial estimate of the geometry, even without odometry or in multiple robot scenarios. This linear estimate can then, if desired, be fine-tuned using 2D bundle adjustment. The algorithm substantially enlarges the scope in which non-linear batch-type SLAM algorithms can be applied. The method is applicable when at least seven landmarks are seen from three different vantage points, whether by one robot that moves over time or by multiple robots that observe a set of common landmarks.
Frank Dellaert's Homepage