
Tommy Tompson from AI and Games gave an in-depth look at the Ubisoft’s team approach to creating a realistic behaviour of NPCs in the Watch Dogs 2 game.
One of the unique features of the second title of the franchise is allowing players dive into the game’s world and lives of other people, make it feel like the real one by building unpredictable human-like reactions of NPCs to a made-up situations in the game. For these purposes, the development team built a Watch Dogs 2 Civilian AI system using Spreadsheets.
The system consists of two parts: a reaction system with hundred of rules written by designers and an attractor system that helps create events for NPCs to react to. Since the reaction system, also known as the reaction matrix, is quite complicated and contains a big amount of rules, the team used a spreadsheet to store all the data. This way, designers and developers can add any changes to it, adjust when needed, and add to the game a completed version.
Attractor systems are placed manually by designers and create situations where the character reacts a certain way. These attractors are also called provocative agents. Dynamic attractors, on the other hand, are generated runtime and create situations like small talks, bumping into each other, and so on.
Find out more details on the systems in the video or in the written article on Gamasutra.
Don’t forget to check Tommy’s YouTube channel where he constantly shares videos with discussions of implementing AI in game development.
You can learn more and get the tool for $20 on Gumroad. Don't forget to discuss it in the comments. make sure to join our new Telegram channel, our Discord, follow us on Instagram and Twitter, where we are sharing breakdowns, latest news, awesome artworks, and more.