Reproduce “RoGSW4RLD: Feed-Forward 4D Gaussian Lifting for Robot World Model Rollouts” 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
Jin Hyun Kim, Min Young Kim, Soohwan Song, Daekyum Kim
RoGSW4RLD reconstructs synchronized robot-camera video as a shared metric 4D Gaussian scene that can be rendered at viewpoints and times. It jointly processes calibrated camera streams and uses robot geometry and kinematics to anchor the robot. On 256 held-out DROID episodes, it improves novel-view image similarity, stereo-referenced depth, and robot displacement error over MoVieS, under calibrated, known-robot conditions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.