Fully autonomous demo: Pick up and put away any piece of clothing.
97% success rate on unseen garments.
Stack:
SLAM + JEV to locate the clothes.
Astra + PI0.5 + collision detector to grasp them.
SLAM + motion replay to put them away.
Average end-to-end time per garment: 2 minutes.
@ClifSeemorris quite nothing, we just send the gripper image to verify the alignement before going down.
even sure that we could use lower models but with this one the full pipeline is at 97% success
Will GPT-6 Astra kill robotics startups? Yes. And no.
Since Astra was launched, 10 robotics researchers around me either cancel their projects or completely reconsider it. But I think the discussion now is sentiment more than fundamentals.
In this video, I covered:
1) what exactly does Astra change
2) why LLM works like a doom
3) Can and Can't for LLM for robotics
4) Emerging moats and opportunities for robotics startup
Open to any discussion in the comment.
Had a great time on @cheddar talking about the future of home robotics and how we're making robots accessible to everyone.
Thanks @Jetsetjournalists for having me!
https://t.co/m9IZvMblfm
@TrunkedMammal That makes sense! Out of curiosity, what specific tasks would you want it to handle before youβd consider getting one? Weβre deciding what capabilities to focus on next, so that would be genuinely useful feedback.