Distributed AI Experiment

[DigitalLibrarian]

Try distributing the agent ai, by including the command queue at the agent level, and define ai commands to do things like compute and en-queue the best possible combat move, or navigational move. You can distribute the ai into more easily testable chunks to accomplish larger goals more re-actively by using the command queue.

Once we have a fantastic search algorithm for finding command sequences, we can parameterize it to perform in different ways. Imagine things like parameterized combat skills and other ai variations.

[DigitalLibrarian]

This is really describing how we can strap decision trees onto the existing infrastructure.