Scan-SLAM: Recursive Mapping and Localisation with Arbitrary-Shaped Landmarks

Tim Bailey & Juan Nieto

Scan-SLAM is a simultaneous localisation and mapping algorithm that combines scan-matching methods with recursive estimation of landmark locations (using an EKF or other Bayesian filter). The scan-matching capability allows landmarks with arbitrary shapes to be modelled directly by sensed data and tracked within a conventional filter framework. This paper presents the essential Scan-SLAM algorithm, and implementation details for application with scanning range-laser sensing: segmentation, alignment, covariance estimation, data association, and landmark model refinement.