I’ve been away from the Nano project for a few months now and I’ve instead been spending my programming time learning Unity. It’s actually pretty easy to use, and after watching a few of the tutorials on Unity’s site I was good to go and started off on my new learning project, which I’ve been calling My Critters. This is an awful name and isn’t a name at all really, it’s just what I say when I tell my girlfriend what I spent my day doing.
Her: “Have you been playing video games all day?”
Me: “No! There were a few hours that I worked on my critters… AND I showered!”
Evolution Simulator Project
In my case, I’m trying to make something I’ve been calling an evolution simulator. The idea is that when you start the simulation a few critters (aka spheres) and some food (cubes) are randomly dropped into a small flat world. The critters have the ability to detect the food via sight (raycasts) and then can move to the food (using physics and impulses) and eat the food. Now, every action the critters do burns energy, some actions burn more than others (moving forward burns more energy than turning) and simply existing slowly consumes energy, and when you run out of energy you die. In the end, critters try to get to food to stay alive, they will mate to make more critters and spread their genes.
This isn’t how it all works right now, currently I only have 1 critter who doesn’t burn energy at all and just poots his way around the world eating everything he sees, sweet life. But all that other stuff is coming and I don’t think it’ll be too difficult to get those basics in place. (In before the post where I talk about that stuff turning out to be really difficult.)
The Hard Part
The hard part comes when the critters aren’t told to do any of those things. The goal is to not have the instruction of move to food, eat food as an explicit instruction of the critters, instead the critters genes should determine what the critter does when it detects food, when it comes in contact with it, when it detects another critter, how it explores its environment, etc. This is when I’ll be getting into neural networks and stuff like that.
This whole idea is based off a project I saw years ago called Polyworld. The guy is a terrible presenter but it’s really inspiring the way the creatures in Polyworld develop their own survival strategies. My goal is to create a similar effect in a 3D space, the bigger goal is to then use this type of creature generation in a game of some sort.
As of now, Nano’s development is on hold. I think Nano makes a decent portfolio piece and I’ve still got ideas for it that I want to implement, so the project isn’t canned, just on the back burner.
I’d really like to get some 3D models in the simulator. It’s really early, but I think having some models to look at instead of just geometric shapes would be more interesting to look at. Since I don’t know much about making 3D models, this part will likely require outside help. For me, the next part is to add more critters to the world, allow for mating, and add energy consumption.
Screen shots in next post!