First working prototype of the simulator. This will likely have a fixed map layout, and may not perfectly model the robot's dynamics, but should allow us to begin working on reinforcement learning.