Goal Oriented Action Planning provides a natural way to create dynamic and smart artificial intelligence. Used in games like F.E.A.R. and Fallout 3, it offers natural behavior with great scalability.

Unlike traditional AI which uses convoluted statemachines and tons of transitions, Goal Oriented Action Planning keeps AI simple, natural and scalable.

This is achieved by using actions and goals. Actions are independent processes which regulate themselves, while goals are used as guidelines for behavior. Thanks to goals the GOAP algorithm knows which actions go together to get certain behavior.

How does the algorithm know which actions go together to satisfy a goal? Actions and goals use conditions, which consists of a string and a value. The action 'Attack Ranged' for example has the precondition [ "hasRangedWeapon", true ]. The action 'Equip Ranged Weapon' has the effect [ "hasRangedWeapon", true ]. Thus the algorithm knows to first equip the ranged weapon, and then use it to attack. This leads to natural, dynamic behavior without ever having to code any transitions!

This asset comes with a ready to use framework and a read me thoroughly explaining how Goal Oriented Action Planning works in both theory and practice. A demo scene is included as well, which showcases 4 actions and 2 goals. A read me for the demo scene is included.