Michael Muehlebach
Papers
6
Total Citations
38
H-Index
4
About
Michael Muehlebach is a robotics and control systems researcher whose work sits at the intersection of model predictive control, reinforcement learning, and data-driven methods for dynamic robotic systems. His research addresses one of the field's central challenges: enabling robots to operate both precisely and at high speed while remaining computationally tractable and safe during learning. His most-cited contribution introduces a time-shift-invariant parametrization for model predictive control that significantly reduces computational complexity in real-time robotics applications, garnering 11 citations. Building on this foundation, Muehlebach has pioneered a feedforward-based reinforcement learning framework that mitigates the risk of system destabilization during training — a meaningful safety advance for physical robot platforms. His iterative learning control work for pneumatic soft-robotic arms demonstrates how deep learning can generalize beyond the single fixed trajectories of classical ILC approaches. Particularly notable is his sustained focus on robot table tennis as a demanding testbed, producing multiple works on ball trajectory prediction, online learning for ball placement, and hybrid gray-box modeling that blends physics with data. His 2024 tendon-driven robot arm work further pushes the boundary between speed and precision. Across roughly 38 total citations, Muehlebach's portfolio reflects a coherent vision: making intelligent, adaptive robot control practically deployable in fast, real-world environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Data-Efficient Online Learning of Ball Placement in Robot Table Tennis2 citations · 2023
- 6