«
Daemon Pan
Software Development
Simulating Player Squads in an MMO Boss Raid
The goal of my thesis was to make a game that shows how artificial intelligence ( AI ) can cooperate with each other during a Massively Multiplayer Online ( MMO ) Boss fight . MMOs include games like Final Fantasy 14 or World of Warcraft .
To achieve this , I implemented AI pathfinding , AI formation , damage avoidance , and AI cooperation . I included four different types of AI squads : Tanks , Healers , Warriors , and Mages . Each has its own unique skillsets and ultimate skill to handle complex situations . I also included a player-controlled red dragon with four skills to create a dynamic battlefield with dangerous fire zones . I used a heat map to show battlefield information like the AI ’ s strength map , danger map , collision map , and Boss vision map . Then I combined it all together to get a tension map that all AI squads can acquire information from . Based on that information , they will use A * or Flow
Field pathfinding to find their best combat position or retreat path . While pathfinding , they will try to keep in a military squadlike formation and cover low-health teammates . In order to make all AI squads cooperate with each other , I made a hierarchical AI behavior manager , which was inspired by the GOAP architecture . The manager can distribute high-level strategies to each squad that the squads will calculate individually based on their class and status and then pass results to squad leaders and members .
This architecture and system can be used in many video games with group AI like MMORPG ( Massively Multiplayer Online Role-Playing Games ) or RTS ( Real-Time Strategy ) games . This thesis took me one year from designing the AI architecture and skills to implementing the system and the artifact in full . Through it , I learned many AI techniques and gathered insights about designing an AI system .
92 SOFTWARE DEVELOPMENT