«
Neeraj Jairam
Software Development
Pathfinding with Influence Maps Occupancy Maps and Potential Fields
For my thesis , I wanted to study AI and pathfinding . This project uses Influence Maps combined with AI algorithm to generate different paths which are specific to a game world . These paths are used by AI to avoid certain negative areas in a map and go towards positive areas in a map . I also implemented pathfinding using potential fields , which is particularly useful when there are multiple agents trying to find a path in a game . Generating Potential Field path is a one-time setup cost and multiple agents at anytime can refer to this to move to a certain location .
Another area I focused on is using Occupancy Maps to guide AI behavior . Using Occupancy Maps is a probabilitybased approach ; when an AI in game loses track of a player , Occupancy Maps can be used to demonstrate searching behavior in which the AI provides a best guess of where the player could be in the game . I implemented variations where
a single AI tries to find a player using its own Occupancy Map whenever it loses track of the player and another version where multiple AI try to find a player . Even if one of the AI lose track of the player , this gave interesting results and can be applied in a game scenario where multiple agents in a game try to coordinate and find a player in a game .
Finally , I implemented various use cases of Influence Maps and Potential Fields and to demonstrate interesting AI behaviors . The development schedule for this project took five months in a sprintbased approach , where after each sprint is a combination of research and implementation .
In this project , I learned many AI techniques and gained insight into how to think about AI from a programming perspective .
92 SOFTWARE DEVELOPMENT