CO-MAN: Safe data-driven control for human-centric systems

The research project CO-MAN aims to develop a novel framework for user-adaptive data-driven control with performance guarantees in order to address the scientific challenges of high uncertainty and individual user requirements.

Motivation

It is important for advances in technology to support human activities and interactions in the areas of healthcare, mobility and infrastructure systems. For instance, making healthcare more human requires digital interfaces to allow for more human interactions with the system. This is the goal of human-centric systems in which the human is both an element of the control system and a design criterion. The EU-funded CO-MAN project will develop a framework for user-adaptive data-driven control with performance guarantees. The biggest challenge will be to merge probabilistic non-parametric modelling techniques from statistical learning theory with novel risk-aware control methodologies while including active user modelling. The game changer is the current push towards reliable machine learning with novel results on theoretical bounds for learning behaviour.

Research topics

• Identification and Control with Gaussian Processes
• Optimal Learning Control based on Gaussian Processes
• Intention estimation during physical Human-Robot-Interaction
• Estimation of human arm impedance for motor behavior models
• Embodiment under autonomous control
• Shared control for human-robot team interaction
• Neuroscientific models of Human-Human and Human-Robot Interaction
• Adaptive torque observer for human-exoskeleton interaction

Links

TUM team members

• Sandra Hirche (principal investigator)
• Satoshi Endo
• Alexandre Capone
• Neha Das
• Giulio Evangelisti
• Armin Lederer
• Hossein Kaviani rad
• Hatem Sultan (project manager)