Zero-shot pick-and-place trained entirely in simulation! 👾 @sudo_robotics held their first public demo showing a mobile humanoid robot doing zero-shot pick-and-place at ICRA. The robot foundation model was trained entirely in simulation, no real-world data, no teleoperation. Pick and place is the fundamental skill. Until you solve it for any item, even items the robot has never seen, you won't solve downstream robotics problems. Let’s have a look at the design: Mobile base with vertical mobility (Z-axis), seven degrees of freedom per arm, three RGB cameras (wrists eyes). Two-finger gripper in this demo but gripper-agnostic for different use cases. Trained under diverse lighting scenarios. Picks up anything. Audience members tested it with items from their pockets, keys, wallets, whatever. It has picked up straws and even my AirPods case 😁 For now this is a platform for developers and researchers. Demo centers opening in Shanghai, Zurich, and Mountain View. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com

We’re live at Booth 638 at CVPR. Come test the robot yourself and see how simulation is helping unlock more capable, general-purpose robotic systems.

Our robot is live at ICRA! Come visit Booth 171 to see our zero-shot pick-and-place demo in action. Bring an item and test it yourself.

A first look at what we’re building at Sudo AI. Introducing #sudo R1. We train robots entirely in simulation, then deploy them directly to hardware - no data collection, per-object tuning, or calibration. In this first release shared in our blog, #sudo R1 runs 60 minutes uncut and reliably picks objects it has never seen before: glass, fabric, metal, irregular shapes - under changing lighting, clutter, and physical interference. Simulation-trained systems rarely hold up in practice. We think that’s starting to change. If you’re building in robotics or thinking about where this is going, follow along at sudo.ai