Our group develops machine learning techniques that include complex physical simulations of the motion of a human body.
Humans interact with the physical world. Exterior forces (e.g. ground reaction forces while walking) act on the human body which in turn require the brain to counteract or interact with them to produce a desired motion. To achieve this, muscles are triggered to create forces. To properly analyse a motion, which is a complex combination of these interior and exterior forces, new approaches are required to calculate these forces.
In contrast to pure human motion capture, interior forces in the body are not measurable by measurement devices. We call the task of estimating these forces Dynamics Capture.