Digital Human Model as Dynamic Manikin

BMBF Project DYMARA: Dynamic Manikin Modeling Skeletal Musculature

In the joint project DYMARA we are developing an innovative digital manikin with detailed modeling of skeletal muscles and fast numerical algorithms.

With this manikin it should be possible to integrate humans into their working environment in an optimal way and to avoid fatigue, illnesses and accidents at the workplace. In addition to these ergonomic aspects, the human model is also used for therapy planning in the muscular area and for the design of prostheses and orthoses.

In order to capture the dynamics of the musculoskeletal system with sufficient accuracy, we follow a modelling approach based on the method of mechanical multi-body systems (MBS). Such models are inspired by robotics and are already used in many biomechanical applications. However, the modelling of the musculature still poses a great challenge, in particular the computing time and the consideration of anatomical and physiological conditions pose challenges.

Our Focus and Tasks in the Project

This is where our project comes in: A new one-dimensional continuum model to be developed, which realistically describes individual muscle fiber bundles, is to replace the previously common discrete force elements in the MBS model. We combine it with fast, problem-adapted numerical algorithms to calculate motion sequences and control the manikin.

Our thematic focus in DYMARA is the optimal control of the entire human model (or a partial model as an arm including shoulder and hand). In particular, we include improved muscle modelling. For specific task scenarios we generate optimal solutions (and thus motion sequences). Another important aspect of this project is the calculation of optimal action strategies. This requires special models, which are then optimally controlled in conjunction with the overall human model.

In the subproject we are responsible for:

  • the efficient numerical optimal control with classical input variables (joint moments)
  • the development of dynamic action models with approaches from the field of Machine Learning (non-linear (GreyBox) model approaches: neural networks, Gaussian processes, local models, methodical studies, model selection)

Project Partners in Research and Economy

In DYMARA we cooperate with research groups of

together with the partners


Project Duration:

December 2016 to December 2019

The joint project is assigned to the BMBF area Mathematik für Innovationen in Industrie und Dienstleistungen.