IWR Colloquium Winter Semester 2025 / 2026 Embodied Intelligence for Humanoid Robots and Exoskeletons through Optimization and Learning

Human-centred robots, from humanoids to wearable assistive devices, promise major societal benefits but require embodied intelligence to move and interact safely with humans. Their complex, nonlinear dynamics and inherent instability make whole-body motion generation and control challenging. In our research, we develop fundamental understanding of human movement, create human, robot and human-robot interaction models of varying complexity and combine model-based optimization and optimal control with learning approaches. I will present examples of advanced motions in humanoids—walking on varied terrain, balancing, riding a Segway or skateboard, bimanual manipulation, and dancing—as well as studies on assistive robots such as exoskeletons and rollators, highlighting how controller design affects support and user comfort.