Just a thought experiment
- What if I, instead of using resistors in a master/slave bicore, coupled the two with incandescent lamps, and enlarged the cap values to compensate for the fairly low resistance. It would appear, according to the circuit cookbook, that when current is fighting the resistance in a light bulb, and the bulb is trying to radiate the energy as light and heat, the result is non-linear feedback. I don't know, but that sounds like something that would result in chaotic behavior between two oscillators, and that's just fine with me. Maybe light bulbs are the next step towards enhanced robot behavior. And the idea also provides some intriguing aesthetic possibilities...
I may try and use them with my sinusoid neurons. That should be interesting.
I'm also caught up on using a master/master setup rather that a normal master/slave setup. Why? Because in my mind, two master bicores would continuously battle for dominance over each other, and the result would be chaotic in itself, as well as double the independent computational power.
And how about memristors? I haven't heard all too much about their potential use in BEAM. Perhaps if I used them in place of the normal resistor in a neuron, or even my sinusoid neuron, then maybe there's some potential there. Perhaps a very simple form of learning would arise? Perhaps even use memristors in N/2 neurons, in a network that processes feedback from the nerve net and from each other, resulting in a fairly complex form of learning.
Now here's the icing on the cake: I want to combine all these principles into a single robot. Yes, an N/2 network with each neuron grounded to all the others and to each sinusoid neuron with memristors, coupled through cap/resistor networks to a sinusoid master/master bicore, each bicore coupled to the other through an incandescent lamp. Perhaps even photodetectors to communicate with other similar robots with the incandescent lamps.
That's a goal to set for myself in advance.