CMU releases Pixel: a low-cost wheeled humanoid with co-designed UMI data pipeline for mobile manipulation
ankurhandos · x · 2026-11-04
CMU researchers Tony Tao and Jason Liu released Pixel, a full-stack robot learning project with two papers targeting general-purpose mobile manipulation.
- Hardware: a low-cost wheeled humanoid with human-aligned kinematics and compliant grippers, co-designed with the human data-collection system so camera viewpoints and gripper geometry transfer from demos to deployment
- Data collection: multiple teleoperation and robot-free human collection interfaces built on a UMI pipeline
- Learning: a unified whole-body controller and a common policy pipeline across human and robot demonstrations; a hierarchical stack enables language-conditioned visuomotor control and long-horizon tasks
- Evaluated on dexterous, active-perception, navigation and whole-body coordination behaviors; advised by Ruslan Salakhutdinov and Deepak Pathak
Related event: CMU Open-Sources Pixel, a Low-Cost Full-Stack Mobile Manipulation Robot(2 posts)→
More from Embodied
- Sutton shares RL-trained robot dog preparing for lunar terrain — RichardSSutton · 2026-11-04
- Open-weight LLMs control robots zero-shot: Qwen scores 31.39 on RoboDojo, beating GPT-6 Astra — pabbeel · 2026-11-04
- Apple reportedly beta tests Apple Watch AI feature that analyzes nearby conversations — Polymarket · 2026-11-04
- Boston Dynamics rebuilt Stretch Gen 2 from 100,000 operating hours, cutting 81% of internal cabling — xmercury_one · 2026-11-04
- Robotics funding hits another all-time high as capital keeps pouring in — Fowe · 2026-11-04
- Robot learns mobility and manipulation with just 15 minutes of training data — GordonWetzstein · 2026-11-04