Making a 3-course Cantonese meal with mobile ALOHA robot

twitter.com

3 pointsmodeless3 comments

Note that the video in this tweet is teleoperation, not autonomous. But there's another video here which demonstrates what it can do autonomously, which is still impressive: https://twitter.com/zipengfu/status/1742602881390477771

I do wonder if this could be done by true teleoperation. There are many cues in cooking like sound, smell, texture, taste that aren't possible to transmit well.

Yeah, smell and taste aren't really possible. Sound is easy though. For texture you could do force feedback, though latency would be really hard to compensate for. Latency is more critical for touch than vision.

But teleoperation over the Internet isn't the goal here. The reason they're doing teleoperation is to collect data to train for autonomy. The operator stands directly behind the robot during teleoperation. And the goal is to be able to train with only a few demonstrations.

Reply on news.ycombinator.com