So, Where’s My Robot?

Thoughts on Social Machine Learning

Class with Joe Jones – Robots at the Tipping Point

Joe Jones is one of the key inventors of the Roomba, and this week he gave a lecture in Cynthia’s Human-Robot Interaction class at the Media Lab. He gave an interesting talk, covering a lot of ideas that we don’t usually worry about in research, but that really enabled Roomba to be such a big commercial success. The main points are in this paper of his.

First some trivia:
Fun Fact #1: The very initial robot vacuum prototype was done by Joe Jones at the MIT AI-Lab Olympics in 1989. The “Rug Warrior” pictured above.
Fun Fact #2: Two companies passed on the robot vacuum idea before Joe went to iRobot and convinced them to fund an exploratory project that turned into what is now the roomba!

There were a couple of ideas that Joe talked about that are really key for commercial robots. You have to be able to quickly answer the question “what does your robot do?” Roboticists can easily get caught up in the bells and whistles and super-techno-geekiness of their robot and forget that in the end, to be a successful business venture people have to understand what the robot does and why they would want it. Thus Joe is a big proponent of the single-purpose robot that has an easy to understand utility. “What does your robot do? — It cleans floors”.

This led to a discussion of the “value” of a robot, which can be a difficult question to answer. Essentially it is a function of cost and utility….and the point Joe focused on is that you can increase value by providing the same utility as a lower cost. This is something that is not usually part of robot design, mostly because of the previous issue. Robots are not generally designed to be special purpose, or to have one main utility.

What does this mean for social robots?

I think these ideas are very important for people wanting to build commercial consumer robots right now. This is great advice for an appliance robot like Roomba, but as soon as we want to build a robot that needs to interact with people to do any part of its task, then the “one purpose” idea is out the window. People bring this whole new level of dynamics and complexity that are going to require robots to have the social and cognitive skills that people expect of a social partner. But in the mean time, I hope more companies are able to find some sweet spots in that consumer robot value function!

December 8th, 2006 Posted by | Industry | 4 comments

Robotic Life on NOVA

The folks at NOVA did a profile recently on my advisor, Cynthia Breazeal, you can stream the segment from their website: NOVA Science Now — Profile Cynthia Breazeal.

I missed the spotlight this time around, but you can see some nice professional footage of my furry friend Leonardo. And some fun footage of Cynthia’s kids playing with the Roomba for the first time.

I think it is great that Cynthia does segments like these from time to time. Occasionally, I’ve heard negative comments from people in academia about the fact that the Media Lab tends to do these kinds of mainstream media spots on PBS etc. It’s seen as “not serious.” But I think this kind of outreach is great, and the reason the Media Lab gets a lot of attention is that a lot of the research that happens here is visual, tangible, and generally demo-able. One of the biggest problems facing the US society today is that not enough kids are deciding to study science and pursue scientific careers. This is true across the board, but particularly for young women. Segments like this one on NOVA make science and engineering exciting, accessible, understandable, and inspiring.

December 6th, 2006 Posted by | In the News | no comments