Clodéric Mars and Jérémy Chanut
It is now fairly common to find autonomous human-like characters that are able to navigate in 3D environments, finding paths and avoiding collisions while exhibiting convincing navigation behavior. In the past few years, several major publications have been applied successfully to games: we now have well-tested recipes to generate navigation meshes (nav meshes), compute paths, have pedestrians follow them, and avoid collisions in a convincing way.
However, we still fall short when it comes to group navigation. Like real groups, we want virtual humans to be able to walk down roads with their group of friends. Like real ones, virtual ...