Jonas Umlauft
Papers
9
Total Citations
231
H-Index
9
About
Jonas Umlauft is a leading researcher at the intersection of machine learning and robotics, whose work focuses on developing data-driven control systems that are both intelligent and safe. His core contributions lie in using Gaussian processes (GPs) to create probabilistic models for robot control, enabling systems to learn complex dynamics from limited data while quantifying uncertainty. Umlauft’s seminal 2017 paper on feedback linearization using GPs (68 citations) pioneered a Bayesian nonparametric approach to model-based control, allowing robots to handle unknown dynamics without extensive prior knowledge. He has also advanced cooperative manipulation and human-robot interaction, notably through Dynamic Movement Primitives (34 citations) and Programming by Demonstration (19 citations), where his uncertainty-aware frameworks capture natural human variability. His work on stable model-based control (27 citations) and uniform error bounds for safe control (17 citations) is critical for deploying learning systems in safety-critical applications. With over 230 total citations, Umlauft’s research bridges theoretical rigor and practical deployment, making him a key figure in the push toward autonomous robots that can learn, adapt, and operate safely alongside humans.
Research Focus
Key Achievements
Top Papers
- 1Feedback linearization using Gaussian processes68 citations · 2017
- 2
- 3
- 4
- 5Bayesian uncertainty modeling for programming by demonstration19 citations · 2017
- 6
- 7Learning stochastically stable Gaussian process state–space models16 citations · 2020
- 8
- 9Smart Forgetting for Safe Online Learning with Gaussian Processes11 citations · 2020