Researchgroups in The Netherlands
Intelligent Control and Robotics, Delft University of Technology
The “Intelligent Control and Robotics” group of the Delft Center for Systems and Control focuses on modular, hierarchical control systems that effectively combine state-of-the-art model-based control mechanisms, as a priori information, while simultaneously allowing for adaptation and learning at every level of the robot’s control system – sensing, perception, state-estimation, control, decision-making and planning.
This approach will facilitate a higher degree of autonomy by providing mechanism for coping with unanticipated changes in the process and its environment and will include high level cognitive functions for effective collaboration with humans.
Our research focuses on the advancement of theory, as well as the corresponding control design algorithms and tools. It is our expectation that these developments will have a profound impact not only on robotics, but also on many other related fields.
We have made fundamental contributions to the following areas:
Reinforcement learning (RL) control
data-efficient approximation techniques for continuous-variable reinforcement learning, convergence properties and performance guarantees, high-dimensional problems, use of prior knowledge.
Nonlinear and stochastic estimation
parameter and state estimation techniques for nonlinear, large-scale systems and time-varying systems
analysis and synthesis for the synchronization of multiple cyclic systems with applications to legged locomotion, discrete event systems, hierarchical control with learning mechanisms.
Prof. dr. Robert Babuska
Dr. Gabriel Lopes
For more information, news and vacancy announcement, please, see: robotics.tudelft.nl.