As you say, one of the major thrusts now is to get the robot to model not just itself, but its immediate surroundings. We're also pursuing the idea of having a group of robots work together to model themselves (http://www.cs.uvm.edu/~jbongard/papers/2007_GECCO _Bongard2.pdf).
You raise a good point about motivation. As far as I know, this is an open question in AI: how can you build a robot that will develop its own motivations?
I've used the starfish 'bodyplan', if you like, for a number of previous projects, so there was no scientific or engineering reason for using it. We've simply built up an understanding of the many ways in which this kind of robot can move, and was interested to see what it would come up with. It's lurching way of moving surprised us: we expecting some form of walking, rather than crawling.
It's hard to make predictions about where this technology will go in the future. What I'd like to do is develop a robot that can model itself and its immediate surroundings. Also, I've done some new work on a group of these robots working together to model themselves (http://www.cs.uvm.edu/~jbongard/papers/2007_GECCO _Bongard2.pdf)
Slashdot Mirror
As you say, one of the major thrusts now is to get the robot to model not just itself, but its immediate surroundings. We're also pursuing the idea of having a group of robots work together to model themselves (http://www.cs.uvm.edu/~jbongard/papers/2007_GECCO _Bongard2.pdf).
You raise a good point about motivation. As far as I know, this is an open question in AI: how can you build a robot that will develop its own motivations?
I've used the starfish 'bodyplan', if you like, for a number of previous projects, so there was no scientific or engineering reason for using it. We've simply built up an understanding of the many ways in which this kind of robot can move, and was interested to see what it would come up with. It's lurching way of moving surprised us: we expecting some form of walking, rather than crawling.
O _Bongard2.pdf)
It's hard to make predictions about where this technology will go in the future. What I'd like to do is develop a robot that can model itself and its immediate surroundings. Also, I've done some new work on a group of these robots working together to model themselves (http://www.cs.uvm.edu/~jbongard/papers/2007_GECC
As one of the authors of this work, I'm happy to answer anyone's questions. ...and no, they won't be taking over the world anytime soon.