A* PATH PLANNING FOR SECOND LIFE Obstacle Avoidance in Dynamic Virtual Worlds

3D Virtual Worlds are becoming a new popular medium for a vast variety of so-called “serious games” applications. These serious games tackle problems like education, commerce, tourism etc. and exploit the notion of environment together with participant immersion within this environment in order to provide better user experience. Second Life is currently one the most popular and powerful Virtual Worlds platforms available on the market and many researchers investigate using it for serious games. To provide the participants with automated support for some of the tasks researchers employ Embodied Artificial Intelligence in the form of autonomous virtual agents. One of the problems Second Life presents for virtual agents is the lack of facilities supporting believable locomotion and obstacle avoidance. In this paper we present our solution to this problem by adapting the A* algorithm to supply virtual agents with path planning abilities in Second Life. The algorithm is adapted to deal with a number of implementation problems caused by the distributed architecture (i.e. network latency, packet loss, etc). We also show how A* path planning (normally used in static environments) can be utilized for collision free navigation in dynamic Virtual Worlds.