I don't know of any online teaching material, I think this is because it is a relatively niche and under explored area of research - which is one of main reason

I don't know of any online teaching material, I think this is because it is a relatively niche and under explored area of research - which is one of main reasons I continue to maintain and develop sharpneat. What has been useful for me is just setting up experiments and seeing what happens, and observing the way things go wrong. Typically evolution will find the flaws in any fitness score (metric) that is defined, often resulting in solutions that give good scores, but only because evolution was 'cheating', or rather, exploiting the flaws in the metric. This problem comes up a lot and it is is educational to go through that process and try to think up more robust tasks and ways of measuring success on those tasks. Ideally we want metrics that are continuous, so that we can 'follow the gradient' of success rather than e.g. having a fitness score just go from zero to a perfect score in one step. Also it's been educational seeing how networks evolve. One of the early modifications I made to NEAT was to periodically disable additive mutations (add nodes and connections), so as to strip away any redundant structure that could be slowing down evaluation of each neural net. This came from just seeing the networks grow rapidly, and not always with an fitness score gains to show for it. In summation I don't have a good answer for you, but if you can get interested in setting up an experiment and seeing how evolution tries to solve it all by itself hen maybe you'll get the 'bug' and start to think about where the limits to this approach are, and different ways of overcoming some of the present limits. From there you might want to pick through the occasional research paper to see what others are doing - I've found the NEAT based research to be quite accessible, i.e. e.g. not requiring a a lot of maths knowledge as is the case with deep learning and the like. Although I admit I do follow that world and have had to learn quite a bit of maths to keep up - and I think that's useful to draw ideas from that area. Hope that is of some help anyway.