Reproduce “Memory-Efficient Boundary Map for Large-Scale Occupancy Grid Mapping” 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
Benxu Tang, Yunfan Ren, Yixi Cai, Fanze Kong, Wenyi Liu, Fangcheng Zhu, Longji Yin, Liuyu Shi and 1 more
Researchers at The University of Hong Kong developed a memory-efficient boundary map representation for large-scale 3D occupancy grid mapping, significantly reducing memory consumption by up to 99% compared to existing methods while maintaining O(1) average query time and efficient updates. The system successfully supported autonomous navigation for a memory-constrained MAV in a multi-kilometer underground parking lot.
Key insights
- A low-dimensional 2D boundary representation can implicitly encode the entire 3D occupancy information, leading to substantial memory savings compared to explicit 3D volume representations.
- A hybrid global-local mapping framework, combining a detailed local map with a sparse global boundary map, effectively balances real-time local performance with large-scale memory efficiency.
- Implementing probabilistic log-odds updates and a clamping policy enables robust handling of dynamic environments and sensor noise, crucial for real-world robotic applications.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.