We are developing small, fast, efficient models for robotics.
Our focus is on models that run directly on the robot. We are interested in useful robot behavior under practical limits on compute, memory, and power.
Research interests
- Compact models. Learning perception and control with small models.
- Fast inference. Reducing the time between an observation and an action.
- Efficient deployment. Making the most of limited onboard compute, memory, and power.
Our aim is to make capable robotics models practical to run on embedded hardware.
Contact
For research and collaboration, write to
[email protected].