‘Artificial Intelligence’ is often proclaimed to be one of the most promising developments in machine learning over the past decade. However, despite our optimistic outlook, the underlying algorithms do not possess ‘intelligence’ in the true sense of the word, but instead are extremely adept at identifying non-linear patterns. In this talk I will highlight how we have leveraged such algorithms to discover optimal control laws in extremely complex systems, especially when effective rules may not be evident a-priori. By coupling autonomous control algorithms with high-fidelity simulations of fish swimming, we have been able to demonstrate that locomotion in coordinated groups may lead to energy savings when individual fish interact judiciously with their companions’ unsteady wakes. Furthermore, adopting a reverse engineering approach has allowed us to understand how these optimal autonomous decisions are related to flow mechanics on a fundamental level. In addition to simulations of group swimming, I will highlight recent work where we make use of Artificial Neural Networks to analyze turbulent flow data, with the aim of identifying extreme events in the flow field.
About the Speaker
Siddhartha Verma received his bachelor’s degree in Aerospace Engineering from Georgia Tech in 2009 and a doctorate in Aeronautical Engineering from Caltech in 2014. During his graduate studies, he studied the turbulent transport of low-diffusivity particles using high-fidelity simulations. After completing his PhD, he moved to ETH Zurich as a postdoctoral scholar, where he combined simulations of bioinspired locomotion with optimization and machine learning techniques. These studies led to a better understanding of how fish exert fine-tuned control over flow, in addition to uncovering the complex interplay between fluid mechanics, sensing, and decision-making in coordinated groups. Dr. Verma is currently a faculty member in the Ocean and Mechanical Engineering department at the Florida Atlantic University.
Host: Professor Mattia Gazzola