Workshop on Physics-Informed Learning for Control: Theory and Practice

In recent years, research at the intersection of machine learning and classical engineering domains has grown exponentially. Machine learning is increasingly being utilized to develop novel data-driven approaches for modeling and controlling dynamical systems, which were traditionally dominated by physics-based models and scientific computing solvers. Conversely, principles from engineering and scientific computing are transforming the machine learning landscape by introducing domain-aware methods, moving beyond purely black-box approaches, and incorporating more structure and prior knowledge into model architectures and loss functions.

Physics-informed machine learning leverages knowledge of the physical world to guide the learning processes of machine learning algorithms. By explicitly integrating physical laws, domain expertise, and prior knowledge into the learning framework, physics-informed learning enables control systems to benefit from the flexibility and adaptability of machine learning while maintaining a strong foundation in understanding the underlying dynamics. This integration results in more efficient and trustworthy learning processes, ultimately leading to superior performance, robustness, and interpretability in control applications. The potential applications of physics-informed machine learning in control and optimization are vast and span a variety of fields. In robotics, for example, physics-informed learning can enhance the control of complex manipulators and autonomous agents by considering mechanical constraints, kinematics, and dynamics. Similarly, in power systems and industrial processes, physics-informed learning can optimize control strategies by accounting for physical phenomena such as heat transfer, fluid dynamics, and thermodynamics.

This workshop aims to provide insight into the latest advances in physics-informed machine learning for modeling, control, and optimization, and to provide the audience with hands-on experience to apply these methods to real-world problems. In the morning, we will provide lecture-style presentations on state-of-the-art physics-informed learning concepts for modeling and control. Afterwards, we will offer step-by-step coding tutorials to provide the audience with the knowledge to apply the concepts to real-world problems. Finally, attendees can work on some simple assignments while supervised by the organizers.

The workshop targets an audience from graduate level to experienced theoretical and practically oriented control engineers who aim to improve their knowledge in physics-informed machine learning for control and optimization.