Robert Babuška
Also known as Robert Babuska
Robert Babuška is a control researcher whose work links machine learning with robot control. He earned control engineering degrees at the Czech Technical University in Prague and TU Delft, wrote a book on fuzzy modeling, and coauthored a book on reinforcement learning. At TU Delft, he led the Robotics Institute and later served as main investigator for the Robotics for Industry 4.0 project at the Czech Technical University; in 2023 he coauthored a neural-network method for symbolic regression. As of October 2026, he is a professor in TU Delft’s Learning & Autonomous Control group, with research interests in deep learning and reinforcement learning.
Timeline
-
1990–1997
Degrees in control engineering
He earns an M.Sc. in control engineering from the Czech Technical University in Prague in 1990 and a Ph.D. from TU Delft in 1997.
-
1998
Fuzzy Modeling for Control
In 1998, he publishes Fuzzy Modeling for Control, a book that explains how to construct fuzzy models from process measurements and design nonlinear controllers from them.
-
2010
A guide to reinforcement learning
In 2010, he coauthors Reinforcement Learning and Dynamic Programming Using Function Approximators with Lucian Busoniu, Bart De Schutter, and Damien Ernst. The book presents reinforcement learning and dynamic programming techniques for decision and control problems in complex systems.
-
2013–2016
Directing Delft’s Robotics Institute
TU Delft opens its Robotics Institute in January 2013, and Babuška describes it as a way to coordinate university robotics projects. In July 2016, he is the institute’s scientific director as Delft wins the Amazon Picking Challenge.
-
January 2017–June 2023
Leading a CTU robotics team
In January 2017, Martijn Wisse succeeds him as scientific director of the TU Delft Robotics Institute, and Babuška remains its Associate Director Research. From June 2017 to June 2023, he is the main investigator for CTU’s Robotics for Industry 4.0 project, which develops machine learning for mobile and industrial robotics.
-
2023
Neural networks for symbolic regression
In 2023, he coauthors a neural-network method for symbolic regression that builds physically plausible models from small datasets and prior knowledge about the system.
-
As of October 2026
Professor of learning and control
As of October 2026, he is a professor in TU Delft’s Mechanical Engineering faculty and Learning & Autonomous Control group. His 2026 publications include work on neuro-evolutionary symbolic regression and semantic target search using MAVs.
Sources and credits
Photo: We have no published photo of Robert Babuška yet. Send us a free-licensed photo.
- Invited Speakers :: ECMR 2019
- Books by Robert Babuška
- Fuzzy Modeling for Control
- Reinforcement Learning and Dynamic Programming Using Function Approximators
- TU Delft Robotics Institute
- Delft wins picking contest
- Professor Martijn Wisse leads TU Delft Robotics Institute
- R4I
- Toward Physically Plausible Data-Driven Models: A Novel Neural Network Approach to Symbolic Regression
- Staff