Extended Abstract on The Application of AI Techniques for Automatic Generation of Crowd Scenes

Animation, especially when involving human figures is both a labour-intensive and time-consuming task. Currently animators, using dedicated software can decrease both factors by applying a keyframing approach (the user specifies key frames and the software interpolates between them) but the process is still tedious and expensive. Those problems become especially apparent in the special effects industry, where it is necessary to generate scenes involving many interacting human figures. Most of the work in such scenes must be done manually by skilled professionals because characters still lack autonomy and self-awareness. Similar problems arise in such domains as computer games and simulations of safe evacuation of crowded areas (tube stations, skyscrapers) in the event of fire or other disasters. In this paper we would like to present our FreeWill prototype which addresses these limitations by proposing and implementing an extendable cognitive architecture designed to accommodate goals, actions and knowledge, thus endowing animated characters with some degree of autonomous intelligent behaviour.