Ciro Santilli considered this as the basis for Ciro's 2D reinforcement learning games, but ultimately decided it was a bit too messy. Nice overall though.