Reproduce “TrajWAM: An Efficient Trajectory-Guided World Action Model for Dynamic Robotic 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
Haohan Geng, Sha Zhang, Weishuo Sun, Guanting Ye, Wenhao Yu, Yuang Xie, Qicheng Zeng, Peng Ye and 2 more
TrajWAM predicts future masks of moving targets to guide an imagined future video, then conditions action generation on that representation. It distills video and action generation to one denoising step each. On DOMINO’s 35-task simulation benchmark, it reports 43.54% success and 54.07 manipulation score, exceeding InternVLA-A1.5’s 29.30% and 42.50. Experiments primarily cover independently moving targets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.