Project : siames
Section: Contracts and Grants with Industry
Keywords : Virtual Human, Real-time Animation, Motion Control, Behavior Coordination.
RIAM AVA Motion.
The AVA Motion program is a partnership with the Laboratory of Biomechanics of the University of Rennes 2 and with the Dæsign and Kineo Cam companies. It was supported by the RIAM (French National Network for Research and Innovation on Multimedia) until june 2004. The main objective of this program is to develop a middleware dedicated to real-time virtual humans (cf. figure 29). It is composed of three layers:
Path-planning: it determines the best path for a character to go to a specific place. It is based on robotics algorithms. This part is handled by Kineo Cam company;
Motion control: it handles the motion of the avatar in order to follow the desired path and to make some actions. This is the main part of our contribution to the project;
Reactive behaviors: it manages the behavior of the actor. It is the top-most layer of this project. It makes the decisions and consequently sets the parameters of the two previous modules as well as their activation and deactivation. This part is made by the end-user partner: the Dæsign company. In order to help them handling several resources shared by many actions and characters, we have integrated HPTS++ inside their behavioral module.
During this project, the MKM (Manageable Kinematics Motions) library was created (cf. 5.3). In addition to this animation engine, we have developped two models: a walking model (cf. figure 30) and a grasping model. They use a reference motion and adapt it according to high level parameters.
MKM uses motions for the human animation. These motions can come from captured or synthetized motions. For the motions created by the infographists, we have developped a 3DSMax plugin (cf. figure 31). It automatically saves a motion to the s4d file format. This is the internal format of the motions. It allows an automatic adaptation to a different morphology and to external constraints such as the environment. MKM can also directly read BVH files, a widely used file format.
Finally, in order to use MKM in a context of productivity, we have created the s4DMaker software (cf. figure 32). It allows to create the s4d motion files from captured or synthetized motions. It also handles additional constraints such as the sequence of support phases which is used during the synchronization process.