In computer and video games, animated films, and training simulations, how autonomous agents--for example, a soldier in a war video game--move and navigate in virtual crowds is a relevant issue. Virtual environment crowd simulation is challenging with respect to the need to have both rich and meaningful behavior control and minimal computational cost. Patil et al. present research that represents progress toward this end. This paper is written for a highly technical audience with some mathematical and programming background. However, since it develops an understanding of what goes into creating realistic group movement in these virtual settings, the topic will also interest casual video game players and people who watch animated films.

What sets this work apart from previous research is the use of guidance fields that are specified by the user either through the sketching of a path into the scene with an authoring tool or by importing video of crowd motion flow fields. A user defines the guidance fields, and an aggregate of all the guidance fields in the environment is used to calculate a smooth, goal-directed navigation field. The proposed algorithm works with all collision avoidance scenarios, and it also provides real-time feedback to provide continuous aid in directing the crowd simulation. The results of the research show that navigation fields can be computed in real time regardless of the complexity of the environment.

This paper on using navigation fields to direct crowd simulations is relevant and has significance in many computing realms where virtual worlds are created either for immersion to escape the physical world or to better understand it.