|I left the population running overnight and things seem to be going well. I like the way they waddle as they move. I'm considering uploading videos in addition to pictures.|
The neural network is work well enough for now. My next goals are to re-enable eating, food spawning, energy use, dying, aging, and saving. The basic idea is to re-enable the evolutionary stuff and the ability to save and load. I'm a big believer in working with a population that persists over time and making changes at a rate at the population can adapt to.
I'd love to implement a humanoid body for the robots but I have to get some basic results first. My goal has been to extend work in evolutionary language by getting the population to develop language by only adjusting embodiments and environments rather than through artificial selection or artificial situations. My hunch is that the former will lead to open-ended evolution while the later, not so much.
First up, re-enable robot eating.
Every now and then I come across work that sounds very similar to mine. What I've found among other things, is that they often used abstract models, or don't aim to move towards human like embodiments and environments. I do get a pang of fear which I've heard researchers get whenever they come across research that sounds similar to their own.
+ Move neuron activation code to inside class Neuron
+ Move synapse adaptation code to inside class Synapse
+ Temporarily move neural network test code into class NeuralNetwork
+ Separate action condition code from action execution code