Pixel: A Co-Designed Full-Stack Robot System for Learning General-Purpose Mobile Manipulation
Pixel: A Co-Designed Full-Stack Robot System for Learning General-Purpose Mobile Manipulation
This paper introduces Pixel, an integrated robotics system designed for general-purpose mobile manipulation. The system combines a low-cost wheeled humanoid with human-aligned kinematics and compliant grippers, alongside multiple teleoperation and robot-free data collection interfaces. It features a unified whole-body controller and a common policy learning pipeline that processes both human and robot demonstrations. A hierarchical learning stack enables language-conditioned visuomotor control and long-horizon task execution. The authors evaluate the system across hardware, data collection, and policy learning, demonstrating capabilities in dexterity, active perception, navigation, and whole-body coordination. Pixel is positioned as an extensible foundation for studying the co-design of robot hardware, data, control, and learning.
The authors present Pixel, a co-designed full-stack system integrating low-cost wheeled humanoid hardware with compliant grippers, human-aligned kinematics, and a unified whole-body controller to enable general-purpose mobile manipulation through combined teleoperation and robot-free data collection.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- code: not_reported
- weights: not_reported
- dataset: not_reported
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
task execution. We evaluate the system across hardware, data collection, and policy learning, and demonstrate mobile-manipulation behaviors requiring dexterity, active perception, navigation, and whole-body coordination. We hope that Pixel provides an integrated and extensible foundation for studying how robot hardware, data, control, and learning can be designed together for robot learning. More results at https://
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org