Daimon Robotics and Galbot released RobOmni around June 5, 2026, presenting it at ICRA 2026 as 'the first omni-modal evaluation benchmark including tactile sensing for physical interaction.' It scores contact-rich manipulation — grasping, placement, precision insertion, assembly — using high-resolution fingertip touch, wrist RGB, gripper state, end-effector trajectories and external cameras.
RobOmni runs on NVIDIA Isaac Sim and includes a digital twin of Daimon's DM-TacClaw gripper, measuring not just success rate but efficiency, dexterity and failure events like slips, jams, collisions and retries. It is vendor-released; an independent peer-reviewed paper was not found, so performance claims should be treated as company-stated.
Vision benchmarks are mature; touch is not, yet manipulation lives or dies on contact. A shared tactile benchmark, if adopted, lets the field compare dexterity honestly instead of on cherry-picked demos — a small but real step toward robots that can reliably handle the physical world.
Key Facts
- RobOmni: Daimon Robotics + Galbot; ~June 5, 2026, shown at ICRA 2026
- Billed as first omni-modal benchmark to include tactile sensing
- Modalities: fingertip touch, wrist RGB, gripper state, trajectories, cameras
- Runs on NVIDIA Isaac Sim; includes a DM-TacClaw digital twin
- Vendor-released; performance claims company-stated
Frequently Asked
What is RobOmni?
A benchmark from Daimon Robotics and Galbot, released around June 5, 2026 and shown at ICRA 2026, billed as the first omni-modal evaluation benchmark to include tactile sensing for contact-rich manipulation.
What does RobOmni measure?
Contact-rich tasks such as grasping, placement, precision insertion and assembly, scoring success rate, efficiency, dexterity and failure events using touch, vision, gripper state and trajectories.
How solid is the evidence?
RobOmni is vendor-released and runs on NVIDIA Isaac Sim; an independent peer-reviewed paper was not found, so its performance claims should be treated as company-stated.