Navigation of intelligent characters in complex 3D synthetic environments in real-time applications

The problem of intelligent agent navigation in unknown and dynamic synthetic environments is a subject of a number of recent AI researches. Even though many solutions have been proposed to address it, ever-growing complexity of virtual environments and development of sophisticated characters make it necessary to further elaborate computational models used to control character behavior making it more robust and believable In this paper we describe our technique for navigation of intelligent characters with significantly different locomotion capabilities in complex virtual environments. The approach is based on the use of so-called accessibility graphs. We depict a general scheme applicable to control significantly different characters in a similar fashion even in the most challenging circumstances. Then, the description of different behavior pattern creation is given within the framework of the model proposed. The technology outlined in the paper is perfectly suited for the use in modern 3D video games.