Singapore unicorn Sharpa Robotics put the most sensor-dense robotic hands on a wheeled humanoid called North. SharpaWave: 22 active degrees of freedom, 1,000+ tactile pixels per fingertip, sub-millimeter resolution, 0.005 N sensitivity, miniature camera inside each fingertip. 30 N grip, digits cycle four times per second. Modular fingers — swap one without replacing the hand.
At CES 2026, North played ping-pong at 0.02-second reaction, framed photos at 2 mm precision, built a paper windmill through 30+ autonomous steps. CES Innovation Award winner.
NVIDIA GR00T: 54% Success Rate Jump
NVIDIA’s GEAR Lab transferred GR00T policies (pre-trained on 20,000+ hours of human video) to SharpaWave-equipped robots. Assembling model cars, operating syringes, sorting cards — success rate up 54%.
Developer Stack: Isaac Lab, ROS2, SharpaPilot
SharpaWave plugs into NVIDIA Isaac Lab, Isaac Sim, and OSMO for cloud orchestration. ROS2 built in. Sharpa’s SharpaPilot app also supports PyBullet and MuJoCo. MANUS glove teleoperation for human demos. Full sim-to-real pipeline, wired.
Mass production started April 2026. If you are building on NVIDIA’s physical AI stack or working on dexterous humanoid manipulation, worth watching.
You Might Also Like
- Anvil Robotics Openarm the 5000 Dual arm Robot kit Already Shipping to Nvidia
- Zeroth Robotics m1 w1 a 2899 Humanoid and a 4999 Cargo bot Ship This Week
- Ricursive Intelligence Just Raised 335m in 4 Months and i Think Nvidia Should be Paying Attention
- Nvidia Personaplex Just Made Every Voice ai Pipeline Feel Obsolete
- Nvidia Dreamdojo Teaching Robots to Think by Watching 44000 Hours of us

Leave a comment