The latest thing is a
swanee wistle connected to a microphone via a computer and some
servos. Primarily inspired by conversations with Roger
Moore involving beer, this device is part of an experimental
set-up looking at mimicry. Can this device, with two degrees
of freedom (pitch & volume) recognizably mimic human
speech? The model is based on Roger's PRESENCE model (PREdictive
SENsorimotor Control and Emulation) of cognition.
It should also provide a demonstration of the complexities of human
interaction and turn-taking in my SALT-E model of attention published
in this paper
this paper which appeared in a workshop
on Companionable Dialogue Systems at ACL in 2010.
Yes it is a walking frame and a pair of crutches, with some pneumatics.
When people ask me why, I say that it is just art, and hence needs no
explanation. Here is the movie... And
some things that distinguish it from other robots:
- 3 weeks to make - and one week of that was trying to get a 8bit C
compiler to run :-/
- The movement is dynamic. The first time I ran it, it threw itself
on the ground. Fortunately I'd read "fast, cheap, and out of control"
and so it was wearing padding at the time. Most robot builders would
simply slow down the action but I wanted that sense of energy that one sees
in the movie. The solution was sensors that prevent the crutches moving
forward when the back legs are off the ground.