Good Feature Matching in BA-SLAM

Abstract: Active map-to-frame matching using SLAM condition scoring is proposed for balancing time cost and accuracy in indirect, BA-SLAM. Exploiting the submodularity property of pose optimization condition scoring leads to an algorithm for deciding when to employ map-to-frame matching, how many points to select, and how to prioritize the data association to best benefit pose estimation outcomes. When applied to ORB-SLAM, the algorithmic modification lowers run-time costs without impacting localization performance.

This work stemmed from the good features selection strategy, but rather than actively filter the points for pose-optimization after data association, the method aims to do so before data association. The map-to-frame data association step is fairly costly in ORB-SLAM as it grows with the size of the map.

Though implemented for ORB-SLAM, the technique is general and can be applied to most any feature-based, BA-SLAM method employing a local pose optimization step for the fast front-end thread.