Reproduce “ObjDream: Guiding Vision-Language-Action Models with Object-Level Understanding and Future State Prediction” 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

Huailiang Ma, Yuxuan Li, Ziyi Zhao, Xinkai Wang, Mingyu Li, Xinyu Guo, Lifeng Zhu, Aiguo Song and 2 more

ObjDream gives robot policies object-level slots that separate persistent identity from changing object state. During training, it predicts future object states conditioned on action chunks, while directly generating actions at inference. On LIBERO-Plus, it scores 84.4% and ranks second overall. Across four real-robot tasks, its out-of-distribution success rate is 55% versus 53% for π0.5, a comparable result in these evaluations.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.