From the oldie-but-goodie department, this fun program uses a genetic algorithm to drive car thingies across a bumpy terrain. Change parameters. Watch the cars go. See how far the winner travels before crashing.
The code is available on GitHub.
In case you’re unfamiliar, a genetic algorithm creates mutations in a population of objects or systems. Those that perform better move on to the next generation. The algorithm keeps going until you get an optimized point. In this case, the algorithm tries to optimize travel distance.
See also evolving floor plans and an optimized brewery road trip. [via kottke]
