Bundle adjustment revisited for SLAM with RGBD sensors

Kathia Melbouci (1) , Sylvie Naudet Collette (1) , Vincent Gay-Bellile (1) , Omar Ait-Aider (2) , Mathieu Carrier (1) , Michel Dhome (2)

1 LVIC - Laboratoire Vision et Ingénierie des Contenus
2 IP - Institut Pascal

We present a method of using depth information provided by an RGB-D sensor, for visual simultaneous localization and mapping (SLAM), in order to improve its accuracy. We present a constraint bundle adjustment which allows to easily combine depth and visual data in cost function entirely expressed in pixel. The proposed approach is evaluated on a public benchmark dataset and compared to the state of art methods.

Keywords: SLAM Simultaneous Localisation and Mapping; Kinect Sensor; RGBD-SLAM; Local bundle adjustment

Domains: Engineering Sciences [physics]; Signal and Image processing

Kathia Melbouci, Sylvie Naudet Collette, Vincent Gay-Bellile, Omar Ait-Aider, Mathieu Carrier, et al.. Bundle adjustment revisited for SLAM with RGBD sensors. 2015 14th IAPR International Conference on Machine Vision Applications (MVA), May 2015, Tokyo, Japan. ⟨10.1109/MVA.2015.7153159⟩. ⟨cea-01830455⟩