Authors: Shannon Pace
There exist a large number of steering behaviour algorithms for holonomic bodies. Many yield a vector steering force with respect to an explicit target. A smaller group combine forces from less intuitive sources to generate emergent individual and group behaviour. The latter typically appears more natural, intelligent and is less sensitive to the environment state. Even so, they enjoy only niche application in games. Applying emergent solutions to more ubiquitous problems such as collision avoidance can make the positive aspects such behaviours mode visible to the audience. This paper introduces an emergent steering algorithm designed to minimise collisions using artificial potential field (APF) methods. Issues typically facing APFs are prevented by a novel, dynamic field perturbation. It is demonstrated to overcome a number of problematic cases. Applications for collision avoidance in game media are ecplored. Implementation considerations and the alogrithm's role within a complete navigation system are also discussed.