Biomechanical modelling
To digitally simulate human motions, skeleton models can be used as a template for possible motions. The range of motion in the model is defined by the joints. Our approach makes use of an OpenSim model, a human skeleton model, as a template for defining the anatomically possible motions. However, to be able to generate realistic movement patterns it is not enough to limit the joints to all possible rotational positions. To replicate planned and structured human movement, motion capture data is collected to guide the skeleton model towards realistic human movement patterns. In this project, we focus on dancing.
The motion capture data is collected as coordinate readings of multiple markers attached to the subject and as a time series, these coordinate readings can express how the body moves in the room (3D). To replicate this movement with our skeleton model we make use of Inverse kinematics. Here, an optimization tries to minimize the distance between recorded markers and the digital markers on our skeleton model, resulting in the movement of our skeleton. This movement created from the optimization is generic and can be scaled up or down to different skeleton frames (as long as the skeletons are reasonably homogenous). This allows for the motion to be replicable to other skeleton models.
To add another layer of realism an avatar of the subject is created in a game engine and the joint angles from the inverse kinematic are applied to the avatar. This presents us with a digital copy, a digital twin, of the subject that is performing either the same dance as the subject or a set of movements that the subject itself has not performed.
A long-term goal of the project is to expand the digital twin with movements whilst adding external forces so that the torque in each joint can be calculated, of which the required muscle force can be determined. This would make it possible to combine the physical movement with other models included in the digital twin, where the energy expenditure and what impact the exercise has could be mechanistically described.
In this project, the aim is to create a pipeline where we can replicate a dance digitally, with motion capture, and animate the same motion to another person. In this way, any motion executed by a professional could be transferred to an avatar of a beginner and act as a visual aid to this person. The hypothesis is that increase motivation and self-efficacy by using this procedure can gain compliance in health-related treatments.