PhD seminar - Michael John Tribou

Candidate

Michael John Tribou

Title

Scale Recovery in Multicamera Cluster SLAM with Non-overlapping Fields of View

Supervisors

Wang, David W. and Waslander, Steve (Mechanical Engineering)

Abstract

The solution to the Simultaneous Localization and Mapping (SLAM) problem using calibrated multicamera clusters is studied. Under general motion conditions, the system is able to accurately track up-to-date relative motion, including scale, between the camera cluster and the (free-moving) completely unknown target object or environment using only image measurements from the set of perspective cameras. The cameras within the cluster may be arranged in any configuration, including those where there is no spatial overlap in their fields-of-view and traditional stereo vision techniques cannot be used to determine scale. A degeneracy analysis for a cluster composed of two cameras with non-overlapping fields-of-view is presented identifying the set of relative motions of the cluster and target leading to degenerate solutions of the nonlinear least-squares SLAM optimization.