Reproduce “Beyond Thinking: Imagining in 360$^\circ$ for Humanoid Visual Search” 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
Jingdong Zhang, Yizhou Wang, Zhengzhong Tu, Xin Li, Wenping Wang, Xiaohang Zhan
The system addresses humanoid visual search by pairing an Imaginator that predicts semantic landmarks in seen and unseen parts of a panorama with an Actor that chooses where to turn. Using HVS-3B as Actor, adding the Imaginator raises success from 48.04% to 62.75% on HOS and from 24.12% to 39.38% on HPS, in simulated searches capped at 10 steps.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.