Skip to content
Technical University of Munich

Human-centered Control

Our research advances human‑inspired control technologies that address challenges where traditional control methods fall short, particularly in systems involving human‑in‑the‑loop dynamics. In such systems, classical model‑based control is hindered by the complexity, variability, and probabilistic nature of human behaviour, as well as by the time‑varying properties of physiological responses such as those evoked by functional electrical stimulation (FES). To overcome these challenges, we develop novel concepts for automatic control of stochastic systems, grounded in data‑driven human models with explicit uncertainty representation.

Our approach is inherently interdisciplinary, combining control theory, probabilistic machine learning, and neuroscientific methodologies. We study how the central nervous system modulates movement through both feedforward and feedback pathways, how impedance is regulated for stability and interaction, and how motor intent can be inferred even under noisy sensing conditions. This knowledge is embedded into grey‑box modelling frameworks that integrate physiological insight with data‑driven adaptability, enabling robust, trustworthy control.

The developed models and methods are applied across a range of human–machine interaction and clinical contexts, including physical human–robot interaction (pHRI), hybrid exoskeleton–FES systems that balance actuation between human effort, robotic torque, and neuromuscular stimulation, adaptive FES control informed by EMG, and abnormal motion detection to support rehabilitation feedback. In parallel, we investigate decision‑support and patient‑state estimation methods for clinical applications such as rehabilitation monitoring, fracture‑healing prognosis, and risk assessment, employing interpretable and uncertainty‑aware machine‑learning techniques.

Through this unified research programme, we aim to create adaptive, personalised, and safe control systems that work in synergy with the human user. These systems are designed not only to enhance immediate performance in assistive and rehabilitative technologies but also to promote long‑term recovery, improve user experience, and generate clinically meaningful insights.

Current topics:

Related projects:

To top