Better robots, please.

It has been heartbreaking watching my daughter get excited about robot toys– and all of their fantastic behaviors that she has read about and seen in fictional shows — only to be ultimately disappointed at how lame they are in real life. Please make better robots.

My daughters have always been able to talk to computers. Alexa isn’t perfect, but she tries and can often be satisfying. Most screens are touch-sensitive. Computers have always been able to delight in surprising ways. But robots have been consistent disappointments. Cinderella (that’s what our daughter named our robot vacuum) is pretty good at sucking up dog hair, but that’s about the limit of her talents.

There are little successes. Keepon is an adorable dancing robot. This classic video is 16 years old.

They actually sold a toy based on this research project, and it wasn’t bad. Not quite as agile and satisfying as the video, but it danced.

Unitree is making these dog-like robots. They’re about $2000 – $3000+ (including shipping). They’re pretty good at locomotion, obstacle avoidance, etc. But that’s their main thing. They can’t even follow a person based on vision, they use a little homing device instead. I’m very impressed, and may even get one, but I’m not sure how excited my daughter would be about it.

There are also many robots that are intended for human interaction, but have not won over masses of fans. For example, Jibo raised $70 million — even made it to the cover of Time magazine as one of the best inventions of the year — yet never made it to mass market.

Is it very hard to make a robot for human interaction? A toy that delights a child, and is robust enough in its behavior to interact reliably and in interesting ways? Perhaps insights into intelligence, computation, and behavior are needed to make better robots.

Are there any robots you know of that are pretty good? Let me know.

By the way, I’m on a search committee for a new faculty member in robotics at UCSB. The ad should be coming out soon.