AI FOR GAMES
One of the topics I learnt at AIE was AI for games. For the assignment about this topic, I implemented multiple forms of path-finding and AI behaviors, these include 2D grid and 2D navmesh, along with state machines and behavior trees. Most of the development time went into making a ease-of-use and practical behavior system that utilized behavior tree and blackboards. The maze game I made along side this contained AI survivors trying to escape and a monster who would chase down these survivors.
The behavior system I built was designed to work directly with a behavior tree in mind. Every behavior has the ability to be searched up from the root of tree, along with being cloned. The design also supports a tree existing in multiple different trees by utilizing shared pointers. Multiples tree could contain one instance of, for example, a mobility tree containing path-finding and steering behaviors.
The maze game was built using a 2D grid with built in functionality for efficient path-finding. The grid was a template, which supported arithmetic values specialized by numeric_limits. The grid has multiple path-finding functions, which include AStar and Flow fields, both of which is used by the maze game.
The survivors shared a blackboard, which contained values pointing to the maze and generated flow field. The survivors would use their position to index to flow field to determine which direction to travel in while avoiding blocked cells. The monster has a unique behavior tree for wandering and chasing the closest survivor.
