AvatarPilot : Decoupling one-to-one motions from their semantics with weighted interpolations

Physical restrictions of the real spaces where users are situated<br/>present challenges to remote XR and spatial computing interactions<br/>using avatars. Users may not have space in their physical environment<br/>to duplicate the physical set-up of their collaborators. Still, if<br/>avatars are relocated, one-to-one motions may no longer preserve<br/>meaning. We propose a solution: using weighted interpolations to<br/>guarantee that everybody looks or points at the same objects locally<br/>and remotely. At the same time, this preserves the meaning of<br/>gestures and postures that are not object-directed (i.e., close to the<br/>body). We extend this work to locomotion and direct interactions in<br/>near space, such as grabbing objects, exploring the limits of our social<br/>and scene understanding, and generating more flexible uses for<br/>Inverse Kinematics (IK). We discuss limitations and applications<br/>and open-source the AvatarPilot for general use.