Robot-Assisted Training in Sports
Concurrent feedback strategies facilitate learning of complex motor task such as rowing. The impact of these strategies depends on the subjects’ skill level and on the specific characteristics of the task. Therefore, we design (i) multimodal feedback strategies addressing different phases of learning, and develop (ii) concepts to automatically switch to the most appropriate feedback for a subject.
We assess human performance and motor learning in real-life tasks and apply statistical learning methods to enhance training. We have developed a tendon-based parallel robot to render haptics in our CAVE or to provide haptic guidance. The robot configuration can be adapted to different tasks by changing the position of the deflection units, adapting the motorized winches, and changing the number of actuators. Highly dynamic movements such as tennis as well as power-intense applications like rowing have been elaborated so far.
Augmented Multiphase Learning
In this project, we investigate the effectiveness of augmented feedback modalities to support multiphase learning of real-life, complex tasks. Feedback methods which can be effective for the learning of novice and skilled subjects are studied for various classes of tasks. For this project, M3 rowing simulator is used to render different task characteristics. When appropriate feedback modalities and designs are found for the successful learning for distinct skill levels and task categories, concepts such as virtual trainer can help supporting a novice subject along multiphase learning process of a complex movement.
In this project, we develop a virtual trainer to enhance robot-assisted human motor learning. Robot-assisted training enables a lot of different training conditions and settings in task rendering and augmented feedbacks. Additionally, in robot-assisted training, a lot of kinematic or kinetic data is available to evaluate and analyze performance and training efficiency. Within the virtual trainer project we are designing and testing real-time capable concepts to close the gap from automated evaluation of training to reasoned automated selection or configuration of a training condition.