Responsive action generation by physically-based motion retrieval and adaptation

International audienceResponsive motion generation of avatars who have physical interactions with their environment is a key issue in VR and video games. We present a performance-driven avatar control interface with physically-based motion retrieval. When the interaction between the user-controlled avatar and its environment is going to happen, the avatar has to select the motion clip that satisfies both kinematic and dynamic constraints. A two-steps process is proposed. Firstly, it selects a set of candidate motions according to the performance of the user. Secondly, these candidate motions are further ranked according to their capability to satisfy dynamic constraints such as balance and comfort. The motion associated with the highest score is finally adapted in order to accurately satisfy the kinematic constraints imposed by the virtual world. The experimental results show that it can efficiently control the avatar with an intuitive performance-based interface based on few motion sensors