One of the main achievements in virtual character animation is the production of motions that are natural to the human viewer. This can be achieved by techniques based on the kinematics of an inner skeleton representation. The feedback of the character animation to the viewer usually uses visually appealing models to enforce the perception of a virtual human. When it comes to producing physically or anatomically plausible motions, additional information and calculation is required, based on anatomical models of the human musculoskeletal system. The animation of a plausible avatar in a wellbeing-oriented virtual scenario could greatly benefit from the assessment of the physical activity of the user. Usually this kind of motion analysis is carried on in highly controlled and restrictive environment. To enable the possibility of physical wellbeing assessment in other environments (e.g. at home), methods inspired from biomechanics, motion analysis and virtual character animation have to be developed, along with the investigation of accommodated low-cost capture devices. In order to achieve this, the anatomy and the physical properties of the human body should be considered during the real-time animation. Examples of parameters used for this task are muscle geometries and their Hill-type model, metabolic energy expenditure and overall balance of the virtual character.

In this work package, we will study and adapt computational methods to estimate the physical wellbeing based upon existing musculoskeletal models. Because of our focus on a low level of restriction, we will study the use of low-cost devices (such as Microsoft’s Kinect camera or a set of accelerometers). These methods will produce estimated measures of muscular activity of the user and drive the characters in the virtual world. Physical fatigue and local soft-tissue deformations will be derived from simulating the human motion. As these methods are typically computationally intensive, research is required to determine the proper amount of optimization and offline computation. Finally the translation of the model outputs to the user through avatars will be investigated.