Simulates a moving AI agent at an unknown location in a grid that must guess location based on a series of unsure observations.
In general, the longer the AI observes (with an expected accuracy merely over 50%), the more highly confident its guess.
For every step, every location in the grid receives a probability, so multiple informed guesses can be made.