Sorry for the lite posts recently, I am buried in tasks for my presentations at CEDEC and Ars Electronica, completing version 2.0 (or maybe I should say 1.0) and creating some great new demos with partners technologies from Visage Technologies (high end PC version) and Mobile Mascots (J2ME handset version).
If you are going to be at CEDEC or Ars Electronica I would be happy to meet.
One of the interesting peripheral, perhaps emergent, benefits of blog posting I have found is that it becomes a personal database of thoughts and ideas that I would otherwise just have floating around, half baked, in my head. I am sure having these thoughts readily retrievable and already written up will be of use at some point or another. Which brings me to this post, the second in my “economics of” “mini series”, and is made up of opinions and observations that I have made over the past few months entries. » Read more: The economics of emotion – Animation
There is a piece of common wisdom [read: old wives tale] that says “we only use 10% of our brains” which leads the listener to exclaim “wow, how intelligent we would be if we could only use the rest!”. Of course this is nonsense, we are fully using 110% of our poor overworked brains abilities, the 10% and I believe it is more like less than 5% is the portion of processing that passes through our executive suite, more commonly known as consciousness.
The interesting figure here is not so much how little of our processing requires executive review but how much does not. Which is bringing me, slowly, to the point in the title, that our behavior is layered with only a small percentage being that behavior that most of the AI community would like to simulate. Well, that is the most interesting part, it is also the least understood area of neuroscience and encompasses the most complex streams of processing. Thats why we dont so far have any reasonable simulations of higher level processing. Higher level simulations will not really be feasible without first constructing lower level simulations, like emotion level.
» Read more: Layers of behavior
The Tokyo Game Show is coming around soon, September 16 – 18. While I have found it to be pretty depressing in terms of interesting content (but pretty good in terms of both hosts) and I may not go never the less it is in Tokyo and some of you may be there? If anyone would like to meet let me know (info at emotion dot ai) and I will do whatever I can to accomodate your schedule (evenings are fine too).
An interesting comment with regard to the inherent serial nature of our current hardware and the parallel nature of the brain. Much of my architecture is involved in faking parallelism on a serial machine by for example “versioning” data, i.e. previous and current data and by ensuring that updates are kept synchronous. The current serial machine does offer one redeeming factor and that is speed. As long as we keep our updates in sync and one complete update can be done ahead of human real time then we can simulate behavior without any loss of fidelity in terms of “frame rate”.
What this means is that if, for example, the brain made 10 calculations in response to an input signal and those calculations each took an average of 100 milliseconds and we wanted to simulate this on a serial machine, as long as the serial machine could perform each calculation in less than 10 milliseconds (x10) then we would not, theoretically, notice any difference in behavior.
» Read more: Faking Parallelism
This looks very interesting, Carnigie Mellon developing an “expressive robot” system as a platform along with tools to allow content creators to build content easily. From the specification (i.e. Maya plug in etc) it sounds like they are making a commerical venture of this. The system is called Interbots.
I am not quite sure to make of this but what it seems these researchers are doing is producing changes in skin tone (modulating melonin and hemaglobin) to reflect emotional changes? I would be interested to hear from anyone who saw this in action, Chris?
I seem to be getting a growing number of readers, either I am doing something right or word is out that there is a mad man rambling here! I would encourage you to dig through the previous posts, there is a lot of interesting stuff there. Please feel free to comment, I seem to get very few comments and while I am happy to keep rambling I am also keen to learn from your comments, thanks!
The brain has no “for” loops. Probably. Using them creates overly rigid behavior.
Following from two earlier posts, “IF only” and “Stream Programming” this posts talks about another evolution in the architecture of the Emotion AI Engine in its new incarnation. Having decided to cull away the “IF” statement wherever possible (its not actually an easy thing to do) I am now also avoiding the use of “FOR/WHILE” loops. Again, I found I was doing this without having actually planned it in the design. » Read more: The problem with Loops
Softimage has is about to release a preview of their upcoming product Face Robot and it looks very impressive. We are going to see if we can integrate with this technology to drive it. However, and this is without seeing more detail I would take exception (as usual) with this marketing blurb..
“Face Robot™ uses a groundbreaking new computer model of the soft tissue of the face to mimic the full range of human emotion.”
From what I can see it does no such thing, it mimics muscle contractions and skin deformation but thats it (impressive as that in itself is). I see no indication that it mimics emotion, it seems that the *animator* does the mimicking of emotion by hand but aided greatly by this technology? Any one have more details? » Read more: Softimage Face Robot
© 2010 Emotionally Challenged · Proudly powered by WordPress & Green Park 2 by Cordobo.
Valid XHTML 1.0 Transitional | Valid CSS 3
