So, Where’s My Robot?

Thoughts on Social Machine Learning

Telepresence Robots

There’s been a recent flurry of announcements of telepresence robots.  QB is now available from the CA based Anybots (shipping Fall 2010).  Texai is the WillowGarage platform (not yet for sale).  And Vgo from startup, Vgo Communications, was recently covered in the Boston Globe.

Some visions of these platforms include: telecommuters logging in to interact with their co-workers at the office, or attending a meeting with co-workers in another city, or using it to check out a situation at the factory from the comfort of your office in another location.  While you’re probably not going to buy one for personal use, (QB has a $15K price tag, and Vgo is about $6K) you might just start to see them roaming around your office.

Vgo and Texai look more like laptops on wheels, but QB has a more anthropomorphic presence that I think is nice.  It’s adjustable height lets it get fairly tall, which is necessary for the telecommuter to get any reasonable view of the remote location, and allows it to participate in standing or sitting conversations.   Yet it is small/skinny/light which makes it safe and easy to operate in close proximity to people.   You drive it with the arrow keys on your keyboard, to make it go L,R,forward,back.

One thing I think it missing from all of these telepresence robots, but particularly from an anthropomorphic one like QB is a neck tilt.   For utility purposes, I imagine it would be nice to scan up and down with the camera.  But also for social purposes, it would be nice for the person to trigger head nods (and shakes if we could add another DOF).   If this robot is going to be standing in for people in conversations, it could be awkward if it doesn’t display any backchannel communication.  People will get used to the robot not doing it, but it’s presence in the conversation would be much stronger with a backchannel.  These are the little “yeahs” and “uh-huhs” that people mutter to let the speaker know “I’m with you, keep going.”   Much of this happens over the speech channel, which QB will capture, but we also use body language to communicate this back channel info to the speaker.   Allowing QB to nod would let the telecommuter give both physical and verbal backchannel to the person they are speaking with.

I’m excited to see how people start using their Anybots and Vgos, the water cooler will never be the same!   Here’s a video of QB in action.

June 9th, 2010 Posted by | HRI, Industry | 3 comments