HRIWeb.org

Just wanted to let folks know that I am a member and I now have access. If anyone is at HRI 2008 and reading this, I hope you're having fun. I really miss not being at HRI, and not being able to attend in Amsterdam!Dennis

It's been a while since posting anything. Getting content together for the site has been more difficult than I expected, but an HRI-related seminar this semester at Washington University should generate some interesting discussions which we'll post about here.

Meanwhile, I've been looking at interaction modeling with Bayes nets this summer, and reading a great deal. I've had some success with toy data and simulated Freeze Tag; in the next couple weeks, we'll be gathering some data of real people playing games to work with, and hopefully some more direct interaction data of a robot negotiating a hallway.

The second day of HRI is over, and I'm beat. I planned to do blogging as the conference ran, but instead I'll be waiting a little longer to type up my notes. More to come later.

HRI 2007 is just a few days away, and I'm still short on content here! At least there are a few blog entries and the paper database so far. I hope to draft a few other students from the Young Researcher's Workshop to help blog the conference and maybe start some discussion over in the forum.

I've been thinking about my previous entry a bit, and while I still think it's an interesting idea, the motivation for it seems a bit lacking. Maybe I'll change my mind on that while traveling on Wednesday, but there doesn't seem to be-- relatively, anyway-- that much interesting periodic motion in the world for the robot to focus on. That said, it's something to play with to hone my matlab skills.

While reading Mindblindness, I had some thoughts about the "Intentionality Detector" (ID). The ID depends on a perception mechanism which determines whether an object is an agent with intents, or just an object being acted on by some force. This would be, in itself, an enormously useful tool. I know that there is some work which already handles this part, but it occurred to me there's another level.

There are intelligent agents, but there also might be "simple agents"-- objects/agents which are acting on their own, but are still following some very simple rules. These might be identified by looking for periodic motion. An example would be a machine operating in the same room as the robot-- perhaps a fan running slowly. Furthermore, there might be agents with periodic motion which is not as simple as the motion of a fan, which we expect to act deterministically.