Why can the bots learn maps themselves?
One of the features that set CGF bots apart from most other bots is their tactical understanding of a map. The bots are able to pick reasonably good sniper locations, move up under cover or patrol stealthily. They also are able to predict threat positions if they lose contact with threats. This tactical understanding is essential to the Action Q2 game (due the single shot kill potential, emphasis on cover and stealth, and dark areas). The implementation of this tactical understanding requires: • complete knowledge of the terrain; • so much CPU power that part of the computations need to be performed before starting the game (the results of these computations are stored in the route files and take up to 6 minutes of CPU time on a Celeron300A) – as a result, the CPU load in-game is acceptable. Basically, this means that without a complete route files, the bots are not capable of Action quality combat. Due to the CPU demands, it is simply not feasible to construct the required information
