So, Where’s My Robot?

Thoughts on Social Machine Learning

Social Robots: the ultimate AI “open world”

Just returning from a week in Chicago at the AAAI 2008 conference. Eric Horvitz is currently serving as the President of AAAI and kicked off the conference with a talk that looked at the history of the field and shared his thoughts about what will move us forward.

He challenged the community to work towards AI in the “open world.” Meaning, get the algorithms and systems out of simulated lab settings and toy problems, and in to real world uncertainty, dynamics, and overall messiness.

The last half of his talk posed a number of challenge scenarios for AI that represent “open world” problems that will be good for advancing the field:

  • Robot manipulation, opening door handles, being able to transfer knowledge and experience to a novel door handle.
  • Semantic robot vision challenge
  • DARPA’s grand challenge and urban challenge
  • Predicting highway traffic, learning patterns of human activity
  • Human-computer collaboration, building shared representations to ground collaborative activity
  • Automated receptionist systems

I was pleasantly surprised with the challenge scenarios that Eric chose to highlight, because without saying so directly, his talk made the argument for social robots being an ultimate AI challenge domain. Essentially the two main open world problems he suggested for AI research were (1) robots and (2) human interaction.

I, of course, whole heartedly agree, and hope to see more social robotics research at AAAI!

July 18th, 2008 Posted by | Conferences | no comments

Enter your password to view comments.

No Comments »

No comments yet.

Leave a comment