Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
OmniContact is a research framework for humanoid loco-manipulation. It uses "contact flow" to connect high-level planning, closed-loop skill chaining, failure recovery and low-level execution, so that a humanoid can perform long multi-step tasks such as carrying boxes and pushing suitcases or carts. It comes from Noitom Robotics with the Hong Kong University of Science and Technology, Wuhan University and the University of Hong Kong, and was accepted to CoRL 2026.
The project page presents a dataset of 74,641 motion-capture clips labelled by skill, with paired object data. It also shows generalized meta skills, meta-skill chaining, recovery behaviours, integration with a vision-language model, and motion sequences on a Unitree G1 humanoid. A browser-based MuJoCo (WASM) viewer runs the ONNX policies, and the page includes a baseline comparison. A YouTube creator reported that the system handles tasks running up to about 40 minutes and that code and a dataset were released.
It is a research release for robotics researchers, not a commercial product. The page text provided does not give a repository link, license or model weights, so those details are unconfirmed. The paper is on arXiv (2606.26201).
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Framework using Contact Flow to teach humanoid robots long multi-step loco-manipulation tasks like carrying boxes and pushing carts, running up to about 40 minutes. Code and dataset released.
Closest neighbours by meaning, not just by category.
Model · Dyna Robotics
World-action model from Dyna Robotics pre-trained on over 1 million hours of egocentric human video for robot manipulation.
Research · Tsinghua University
Tsinghua research system in which a humanoid robot learns to play badminton rallies from limited human motion data.
Research · Nvidia
NVIDIA research framework that teaches dexterous robot hands manipulation from human demonstrations via contact wrenches.
Research · BitRobot Foundation
Open dataset of 500+ hours of whole-body teleoperated Unitree G1 humanoid data recorded in real homes.
Research
Research framework adding sole pressure-insole tactile sensing to humanoid locomotion policies, tested on Unitree G1.
Model · Xiaomi
Open-weights Vision-Language-Action robot foundation model from Xiaomi, pre-trained on 100K+ hours of UMI trajectories.