Robot Learning's Big Week Opens in Austin: a Hand-Capture Dataset Drops as CoRL 2026 Fills the JW Marriott
The Conference on Robot Learning runs November 9-12 in downtown Austin, and Monday's hometown headline is a free release: 11 hours of precision hand-tracking data, published to train the next generation of hand-reconstruction AI.
AUSTIN — The Conference on Robot Learning — CoRL 2026, one of the field's main gatherings — opened in Austin on Monday, running November 9-12, and the week's first Austin dateline is a giveaway: Noitom Robotics announced it will openly release HiPHI-Hand, a precision hand motion-capture dataset, on Hugging Face under a research license.
According to the company's GlobeNewswire release, HiPHI-Hand carries sub-millimeter hand ground truth with synchronized first-person video: 11 hours of synchronized dual-fisheye egocentric video, camera extrinsics referenced to the headset's motion sensors, and 3D hand joint positions from optical motion capture. The dataset was used to train NVIDIA's EgoHand, a model that reconstructs 3D hands from a single camera's video, built on NVIDIA's Isaac Capture tools.
The release is not just a download. On the JW Marriott's third floor, booth 1105, from Monday through Wednesday, attendees can put on Noitom's SOL ego-vision headset and watch their own hands reconstructed in 3D in real time by EgoHand running on NVIDIA RTX PRO hardware, the company says — the kind of demonstration that turns a dataset announcement into the conference's hallway conversation.
Hands are one of robotics' hardest problems: grippers and humanoid hands fail where human fingers improvise. An Austin week built around that problem — with its training data given away free — is the local tech story under the conference banners. CoRL 2026 runs through Thursday, November 12.
