INSAIT has a new paper accepted, now at a premier robotics research conference (IROS). The event takes place October 14-18th in Abu Dhabi.
The paper, named “Ternary-type Opacity and Hybrid Odometry for RGB-only NeRF-SLAM” explores improving RGB-only NeRF-SLAM by using ternary-type opacity instead of binary opacity for 3D scenes with opaque surfaces, providing better optimization through volumetric rendering.
SLAM (Simultaneous Localization and Mapping) is crucial for robotics to autonomously navigate and map unknown environments while accurately determining their position. This capability is necessary for obstacle avoidance, autonomous exploration, and real-time adaptation to dynamic environmental changes.
The INSAIT team includes PhD student Asen Nachkov and senior researchers Prof. Luc Van Gool and Dr. Danda Pani Paudel, as well as Junru Lin, a student at the University of Toronto and part of the INSAIT SURF Undergrad Summer Fellowship Program in 2023 during which she worked on the project, as well as Dr. Songyou Peng, research scientist at Google DeepMind.