Object interaction in real-time virtual environments

Marcelo Kallmann
2005
This thesis is about the problem of how to achieve real time virtual environments with autonomous virtual human actors, which can interact with virtual objects in order to achieve a given task. The focus is on interaction with day life objects having some proper functionality and purpose, as for example: automatic doors, general furniture, or a lift. The proposed approach is based on a complete definition and representation for interactive objects. A graphical modeler application was
more » ... y developed in order to define such representation of interactive objects, which are called smart objects. This representation is based on the description of all interaction features: parts, movements, graspable sites, functionalities, etc. In particular, smart objects keep interaction plans for each possible actor-object interaction, detailing all primitive actions that need to be taken by both the object and the actor, in a synchronized way. Regarding the shape representation of objects, a new boundary representation data structure is introduced, providing low storage space requirements together with constant time access to adjacency relations; what is needed by many geometric algorithms. An agent-based simulation environment is also presented with the built-in capability to simulate actor-object interactions, p roviding an automatic actor animation control for interactions with smart objects. The agent common environment (ACE) system is extendible and controllable with interactive Python scripts, and has been used as a system platform for research on behavioral animation. ACE incorporates many new solutions regarding the control of interactive virtual environments, including the interaction with smart objects using virtual reality devices. The approach proposed in this thesis was tested in many different applications, and the results obtained are shown and discussed. -iv -Acknowledgments It is an impossible task to not forget the many people who have contributed, directly or indirectly, to this work. As I could not simply omit such a page, I would like then to express all my thanks: To Prof. Daniel Thalmann and all assistants of LIG for their invaluable support, in particular: Etienne de Sevin, Jean-Sebastien Monzani, Angela Caicedo, and Anthony Guye-Vuilleme, for all the help and motivation with Somod, ACE and Python; Paolo Baerlocher, Christophe Bordeux, Luc Emering, and Ronan Boulic for the many implementations and tunning of all the libraries; Selim Balcisoy, Soraia Musse, Nathalie Farenc, and Elsa Schweiss, for the many work done in collaboration; Tom Molet, Fabien Garat and Serge Rezzonico, for all the help concerning virtual reality devices;
doi:10.5075/epfl-thesis-2347 fatcat:jeo5b2m6inbzlptobxbpetoo7y