← Back to live feed1 story
Robot policies trained in simulation can now operate on physical hardware without manual calibration via Asimov's zero shot sim2real locomotion. The system transfers movement patterns directly to the robot, bypassing the tuning process typically required to bridge the gap between virtual and real environments.
The open source project is designed to assist independent developers and garage robot builders in faster prototyping. Asimov is currently accepting orders for the hardware compatible with the new policy transfer method.
Sign in to suggest edits
Key sources
- SOURCE@asimovinc“You train the policy in simulation. It runs on the robot without additional tuning.”x.com
- SUPPORT@brianroemmele“Policy transfer from simulation in a one shot is a massive game changer for garage robot builders.”x.com
- SUPPORT@sentdex“Asimov is now zero-shot sim2real for locomotion.”x.com