Sunday, January 03, 2010

we're building robotic creatures - we decided to start with the rat

Dan Lockton would probably be interested in this...



The robot is this Bristol Robotics Lab project; both people in the thread at Jamie Zawinski's I saw it in, and everyone I've shown it to, immediately think it looks like a cat.

In fact, in a sense, they do recognise it as a cat - it's roughly the right size and shape, it's in the right place, and if you wave a piece of string in its whiskers it responds much like a cat. But actually, as the project Web site tells us:
The robot was designed to reproduce the behaviour of rats as they use their whiskers to explore their environment. To get a clearer picture of how rats use their whiskers we filmed them using high speed video cameras (500fps) and manually tracked the position of each whisker in the array on a frame by frame basis. Software based automatic tracking is still very much in its infancy though there are a number of groups (including our Sheffield partners) who are now working toward such an application.

The data from this whisker tracking allowed us to quantify the kinematics of whiskers as the rats explored novel environments. From this we found that following a whisker making contact with an object there was a very rapid (~13ms) change in the velocity profile of the ‘whisking’, or movement pattern of the whiskers. We also observed that the rat will tend to move, or orient, its nose toward the exact point of contact.

Our hypotheses were that the rat was trying to optimise the force applied by the whiskers making contact with the object as well as bringing as many addition whiskers as possible, and its nose for smelling, to bear on that point. The orienting behaviour we see as an example of a higher level control loop through the brain, very similar in nature to how we as visually dominant animals rapidly orient, or saccade, the fovea of our eyes toward interesting events detected by our peripheral vision.

To this end we designed our robot to mimic both the low level contact mediated adaptation of the whisker motion pattern and the ability to orient its ‘nose’ towards points in three dimensional space. Designing the physical robot to be capable of mimicking these behaviours allows us to test different computational models of the underlying brain structures which can control it.
So it's a ratbot, and interestingly enough, it's an example of a hardware simulation of a biological phenomenon.

But this is also an example of an interesting design phenomenon; if you want objects to be immediately comprehensible, it helps to use the patterns Dan details here; notably, in this case, similarity, mimicry, and role-playing. Everyone knows how to act around a cat; a robot, not so much. We've been trained to do so by cats. And this part of the project will be unavoidably cattish:

We hope to be able to demonstrate the validity of the proposed brain model by the robot being able to chase an object (perhaps a remote controlled car) moving through its whisker field


Aww. Also, our associations for cat- and also dog-like behaviour stimulate our curiosity towards them, in part because we project it on them. Just as making a new smartphone an interesting object to handle speeds the learning process, a robot that encourages curiosity and interaction towards itself will speed its users' learning process.

I suspect that the response to the Bristol Scratchbot would have been rather different had we been told in advance that it was emulating a rat. (Special note - much of the robot was made on a rapid prototyper.)

1 comment:

Anonymous said...

Nah.

Snake with a fat bum.

kostenloser Counter