Artificial intelligence news.
Straight from the original announcements.

ROBOTERA launches XHAND 1 Pro-U for NVIDIA GR00T 2 demonstrations

November 16, 2026, 11:04By the AI News Bites Editorial TeamSource: PR Newswire
ROBOTERA launches XHAND 1 Pro-U for NVIDIA GR00T 2 demonstrations
Image: PR Newswire

ROBOTERA has commercially launched XHAND 1 Pro-U, a five-finger robotic hand featured in an NVIDIA GR00T 2 demonstration at the Conference on Robot Learning. The company says the hand and NVIDIA GEAR Lab's DexUMI exoskeleton can connect human demonstrations with robot execution.

Key points
  • XHAND 1 Pro-U is a five-finger robotic hand with 16 degrees of freedom.
  • The hand was featured in an NVIDIA GR00T 2 demonstration at CoRL.
  • DexUMI captures hand movements and fingertip force information for model post-training.
  • GR00T 2 was tested on three multi-step tasks with success rates of 60%, 80% and 75%.
  • Commercial orders are open, with an initial limited release of 40 units.

A robotic hand designed for direct skill transfer

XHAND 1 Pro-U has a human-like configuration and a 16-degree-of-freedom mechanical structure closely aligned with NVIDIA GEAR Lab's DexUMI exoskeleton. ROBOTERA says the corresponding joints allow movements to map directly between the exoskeleton and the robotic hand, reducing the need for additional motion conversion and helping preserve details from human demonstrations.

The redesign also focuses on thumb positioning, palm-side tactile sensing and hand articulation. These changes are intended to support natural demonstrations and a range of grasping postures. Together, DexUMI and XHAND 1 Pro-U form what ROBOTERA describes as a demonstration-to-deployment workflow for capturing human skills, post-training a model and executing tasks on a robot.

DexUMI demonstrations train GR00T 2

With the exoskeleton, users demonstrate tasks by manipulating real objects. DexUMI records hand movements and fingertip force information, capturing both how an action is performed and how force is applied during physical contact. Those demonstrations are used to post-train GR00T 2, after which the resulting skills are transferred to a dual-arm robot equipped with two XHAND 1 Pro-U hands.

ROBOTERA says the approach differs from conventional teleoperation because operators do not need to remotely control the target robot during data collection. For the tasks shown, post-training used DexUMI demonstrations without additional task-specific teleoperation or extra robot play data.

Three multi-step tasks tested

The GR00T 2 system was evaluated on preparing banana yogurt, scooping granola and fire chemistry. The last task involved striking a match, lighting a lamp and placing the match in water.

According to ROBOTERA, the system achieved overall success rates of 60%, 80% and 75% on the three tasks, respectively. The demonstrations included tool use, object transfer, lid handling and coordinated two-handed actions. The company says the results show the system executing complex, multi-step tasks through physical interaction.

Commercial orders open with limited initial release

XHAND 1 Pro-U is now available for commercial orders, with an initial limited release of 40 units. ROBOTERA is seeking inquiries from research teams, robotics developers and industry partners working on dexterous manipulation, robot learning and real-world robotic applications.

The launch gives those groups a commercial hardware option designed around the DexUMI workflow shown with GR00T 2. The source supports the use of the hand in research and development applications, while the limited release indicates that the initial availability is restricted to 40 units.

Related news

NVIDIA Unveils Simulation Tools for More Precise Robot Skills
ModelsNovember 9, 2026, 15:00

NVIDIA Unveils Simulation Tools for More Precise Robot Skills

NVIDIA is introducing open robotics technologies at the Conference on Robot Learning in Austin, Texas, aimed at teaching robots delicate tasks such as aligning gears…

Berkeley researchers outline an agentic approach to robot programming
ResearchNovember 9, 2026, 15:00

Berkeley researchers outline an agentic approach to robot programming

UC Berkeley researchers have outlined a robotics approach in which teams of AI agents generate, test and refine robot-control programs in simulation. They call the…

NVIDIA Reports 10x Growth for Open Source 6G Tools
For developersNovember 6, 2026, 17:00

NVIDIA Reports 10x Growth for Open Source 6G Tools

NVIDIA says global developer adoption of its AI Aerial and Sionna software has grown 10x over the past year, while its open source 5G and 6G tools have passed 1 million…