Double window optimisation for constant time visual SLAM

Abstract

We present a novel and general optimisation framework for visual SLAM, which scales for both local, highly accu-rate reconstruction and large-scale motion with long loop closures. We take a two-level approach that combines accu-rate pose-point constraints in the primary region of interest with a stabilising periphery of pose-pose soft constraints. Our algorithm automatically builds a suitable connected graph of keyposes and constraints, dynamically selects in-ner and outer window membership and optimises both si-multaneously. We demonstrate in extensive simulation ex-periments that our method approaches the accuracy of off-line bundle adjustment while maintaining constant-time op-eration, even in the hard case of very loopy monocular cam-era motion. Furthermore, we present a set of real experi-ments for various types of visual sensor and motion, includ-ing large scale SLAM with both monocular and stereo cam-eras, loopy local browsing with either monocular or RGB-D cameras, and dense RGB-D object model building. 1.

Keywords

Affiliated Institutions

• Imperial College London GB
• Universidad de Zaragoza ES

Related Publications