Buzz Zap!!! We were talking about robots after being inspired by this video on you-tube. It shows a robot able to cover a variety of terrains without falling over. Wow!
Being nerdy, we all had our input into the conversation. Here are a few of the things I took from us talking:
--memory and learning are simpler than you might think--Rosenfield uses a constructivist approach to explain how memory is built. His take is that memory is made up of the connections between neurons (rather than just being a retrieval system) and is therefore constantly being created by simultaneous stimuli that create a unique pathway in your brain.
--actually the other way of dealing with learning robots makes sense too. We have instincts which are 'hard-wired' into our mind--like little subroutines that can be 'called' when they're needed
--Lego robots--modular electronic components fused with legos, that could be built lego-style into a circuit--would be pretty cool. Kids could learn how to build working robots, people could make robots based on the need of the day
This morning I was thinking back on all of this and realized that going from simultaneous stimuli like Rosenfield talks about to actual stimulus & response behavior --it's all about the coding. How can you take impressions of the big world and make them digestible for a computer brain to decide what actions to take? What is the streamline-iest way to make a robot learn? Could we make a rubber-band powered robot?
Nerdy nerdy nerdy
As for making the big world digestible, I think the next big wave is going to skip reading research articles and just reading reviews of these articles, because there's too much to go through. That's what I'm doing for now anyway.