Raunhardt, DanielBoulic, Ronan2012-06-122012-06-122012-06-12201110.1002/cav.378https://infoscience.epfl.ch/handle/20.500.14299/81601WOS:000297631000002Despite the large success of games grounded on movement-based interactions the current state of full-body motion capture technologies still prevents the exploitation of precise interactions with complex environments. The first key requirement in the line of work we present here is to ensure a precise spatial correspondence between the user and the Avatar. For that purpose, we build upon our past effort in human postural control with a prioritized inverse kinematics (PIK) framework. One of its key advantages is to ease the dynamic-and priority-based combination of multiple conflicting constraints such as ensuring the balance and reaching a goal. However, its reliance on a linearized approximation of the problem makes it vulnerable to the well-known full extension singularity of the limbs. We address this issue by presenting a new type of 1D analytic constraint that smoothly integrates within the PIK framework under the name of FLEXT constraint (for FLexion-EXTension constraint). We further ease the full-body interaction by combining this new constraint with a recently introduced motion constraint to exploit the data-based synergy of full-body reach motions. The combination of both techniques allows immersive full-body interactions with a small set of active optical marker. Copyright (C) 2010 John Wiley & Sons, Ltd.inverse kinematicslimb singularitypostural controlsynergyextrapolationInverse KinematicsArticulated FiguresPostural ControlMotion CaptureReal-TimeArchitectureAnimationMovementDesigntext::journal::journal article::research article