Reproduce “Map2SLAM: From Streaming Map Prediction to Real-Time SLAM” 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
Jian Gao, Lin-Zhuo Chen, Yihang Chen, Jiagang Chen, Jiacheng Dong, Xun Cao, Nan Xue, Yinghao Xu and 1 more
Map2SLAM turns predictions from the frozen LingBot-Map geometry model into revisable SLAM observations by combining causal tracking with retrieval and geometric checks of revisited frames. Verified loops update an online pose graph, while selected history supports later dense refinement. Across three benchmarks, it reports 42.7–48.3% lower mean ATE than second-best uncalibrated methods, though calibrated DPV-SLAM++ performs better on Oxford Spires.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.