Reproduce “SLAM Dunk: A Recipe for Long-Horizon Egocentric SLAM in the Wild” 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

Jacob Van Meter, Ayush Paul

SLAM Dunk improves offline visual–inertial bundle adjustment by allowing stereo-camera calibration to vary over time and fitting gyroscope scale and axis-misalignment errors. On the held-out LaMAria leaderboard, it exceeded Aria in three of five categories, including Medium (78.9 versus 78.5 CP), but its equally weighted category mean was lower (75.46 versus 75.86). This supports near-parity on the benchmark, not an overall win.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.