A founder answers

Why was the 24-hour napkin test such a breakthrough?

Most robot demos are very brittle — it takes many shots to get even one video that works, and prior works often hit only 70% or 80% success. Dyna shot a 24-hour video of a robot folding nearly 800 napkins non-stop with a 99% success rate. Getting a robot robust enough to sustain a task for a long duration was a technical barrier that hadn't really been solved before their work.

The full answer

JM
Jason Ma · Dyna Robotics
EP 14 · Founder, Dyna Robotics
Show notes ↗

Most robot demos are very brittle — it takes many shots to get even one video that works, and prior works often hit only 70% or 80% success. Dyna shot a 24-hour video of a robot folding nearly 800 napkins non-stop with a 99% success rate. Getting a robot robust enough to sustain a task for a long duration was a technical barrier that hadn't really been solved before their work.

More from this episode

Jason explains why the result was so impressive in the research and robotics community: you've probably seen lots of demos of robots doing cool stuff, but typically those demos are very brittle — "it took many many shots to get even one video that works very well." Getting robots to be very robust and "sustain a long duration of like actually doing a task is a technical barrier that hasn't been really solved before our work," especially for highly dexterous, complex manipulation tasks like folding laundry and napkins. Across a span of a week they shot 24-hour videos of the robot continuously folding napkins non-stop, hundreds folded without much failure. He contrasts this with prior works where you can get a robot to do a laundry-folding demo "but the success rate is often like about 70% or 80%" — meaning if you fold 10 t-shirts you might only succeed eight times. "That's good enough for a demo but it's not good enough for actual real world deployment," because a robot that succeeds 8 out of 10 times "would be a very frustratingly bad robot."