System identification is about constructing models of dynamical systems from experimental data.
System Identification, or the discipline of learning dynamical systems, is an area closely related to cyber-physical systems, as well as real-time big data analytics, and it provides backbone algorithms for digitalization of industry and society. Among others, it is core technology in autonomous systems with applications such as smart buildings, self-driving vehicles, and self-learning robots.
At the Division of Decision and Control Systems, there is a strong group of researchers addressing the problem of data-driven learning of complex dynamical systems. Some of the challenges currently been tackled include:
- Development of fundamental techniques to learn parsimonious models in a statistical and computationally efficient way; here, tools such as regularization play a key role, as they allow to incorporate prior knowledge into the learning process.
- Active and on-line learning, which concern how to improve data-efficiency by actively controlling the excitation of the system in a sequential manner, possibly together with an application specific objective.
- Learning of dynamical networked systems: this is a highly relevant topic, due to the omnipresence of interconnected systems, a field rapidly increasing in importance thanks to the fast development of (wireless) communication technology and the Internet-Of-Things paradigm.
The faculty members of the System Identification Lab at the Division Decision and Control Systems are: