Reproduce “F2A-ObjectNav: An Adaptive Agentic Architecture for Object-Goal Navigation” 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
Seyed Ali Mirghasemi, Alireza Taheri
The authors present F2A-ObjectNav, an object-goal navigation system that uses a rule-based supervisor to coordinate search, candidate validation, memory, recovery, strategy changes, and conditional semantic advice. In 25 episodes per environment, it achieved 84% success and 0.52 SPL in AI2-THOR, versus 40% and 0.22 in RoboTHOR. These descriptive results do not establish an adaptive advantage over a matched fixed system.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.