Reproduce “RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design” 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
Tianxing Chen, Yuran Wang, Mingleyang Li, Yan Qin, Hao Shi, Zixuan Li, Yifan Hu, Yingsheng Zhang and 11 more
The authors introduce RMBench, a simulated benchmark for robotic tasks requiring memory of earlier observations, and Mem-0, a modular policy for testing memory pathways. They classify tasks by the past information needed for decisions and compare eight policies across nine tasks. Mem-0 averages 42.0% success, versus 19.4% for MemoryVLA, but all policies score 0% on Press Button.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.