Reproduce “EvoMem-VLA: State-Evolution Memory for Long-Horizon Robot Manipulation” 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
Yuheng Na, Zhide Zhong, Junjie He, Junfeng Li, Haodong Yan, Jiaan Wang, Jiaguan Zhu, Yangyang Zheng and 2 more
EvoMem-VLA helps robot policies use changes between past and present images for long-horizon manipulation. A learned, source-conditioned delta token is paired with its target state, giving the policy evidence of observed change rather than only remembered snapshots. On RMBench, one jointly trained policy achieves 80.7% success versus EventVLA’s published 67.8%; controlled ablations support this representation’s value within that setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.