Reproduce “KinoGen: Customizable Humanoid Loco-Manipulation References via Kinematically Grounded Video Generation” 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
Qi Wu, Lulu Zhao, Sanghyun Hahn, Rishi Yennu, Emmy Yang, Dingqi Zhang, Tao Pang, Jiuguang Wang and 1 more
KinoGen generates robot-specific loco-manipulation references by rendering the robot, object, and scene, conditioning video generation on two calibrated views, then fitting the generated video to robot and object configurations. This kinematic grounding avoids a human-body proxy and retargeting. Across eight simulated tasks, it preserves 91.16% of annotated hand–object contact frames versus 69.40% for GRAIL, while GRAIL has lower root acceleration.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.