Finite State Machines vs Goal Orientated Action Planning

This project will compare and contrast both FSM and GOAP based AI in a series of tests designed to test performance, realism and dealing with unexpected events. Each system will then be ranked to gauge the benefits of one from the other. At the end of the project there will be a playtest of a game where the AI can switch between FSM and GOAP AI systems, this will allow the player to experience the difference in the AI and gauge the realism and immersion value the different AI allows.

Apologies in advance, this is really just a quick scrap for my notes so I can remember how I did things. If you have any questions though feel free to ask!

Rough Outline

After playing around for a while with the character controller for the Bootcamp demo, I decided that in order to interact with it, instead of re-writing the whole thing I would tie requests into it by using a blackboard system. The player controller asks if the player wants to shoot by default (Fire button pressed?) so I simply told the AI that if it wants to shoot to write in the blackboard (shoot = true;), the Controller then reads in the value and starts to shoot.

For Pathfinding I am using Aron Granberg’s amazing library to generate paths, this allows me to focus more on the AI than the Pathfinding issues.