Reproduce “RAMEN: Real-time Asynchronous Multi-agent Neural Implicit 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
Hongrui Zhao, Boris Ivanovic, Negar Mehr
RAMEN introduces a real-time asynchronous multi-agent neural implicit mapping system that enables robust scene reconstruction despite unreliable communication. This system, demonstrated as the first real-world implementation of its kind, achieved over 91% average completion ratio in simulations under 50% communication success and successfully converged in hardware experiments where synchronous methods failed.
Key insights
- Communication in real-world multi-robot systems is inherently asynchronous and unreliable, necessitating robust approaches beyond synchronous distributed optimization.
- Incorporating uncertainty weighting into decentralized Consensus Alternating Direction Method of Multipliers (C-ADMM) allows agents to effectively leverage outdated or partial information without degrading map quality.
- A simple frequency-based uncertainty quantification method, based on counting non-zero gradient updates for feature grid parameters, can effectively guide consensus in neural implicit mapping.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.