Reproduce “RoMo-GS: Efficient Online 3D Gaussian Splatting for Robot Mapping via Exposure-Integrated Motion Blur” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sangwoo Jung, Di Deng, Jiahui Fu, Laura Herlant, Ayoung Kim, Yafei Hu
RoMo-GS builds online 3D maps from moving robot cameras while accounting for exposure-time blur. It analytically integrates each projected Gaussian’s opacity contribution over exposure, avoiding repeated temporal renders, and prunes low-impact Gaussians while refining camera state and map. In ETH3D, it runs at 3.695 fps versus 1.592 for ten-sample rendering, with slightly lower reported metrics; this comparison uses a pseudo-reference.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.