MAE Department - PG Seminar - Data Science and Computational Intelligence for Air Transportation Systems - from Airliners to Delivery Drones

In an era where affordable, robust sensors generate vast amounts of data, the fusion of this data with physics-based models is radically transforming engineering systems, with the promise of better understanding of system behaviours and tailored prediction of system responses. However, this is especially challenging in air transportation systems due to 1) the computational complexity of modelling these multi-agent multi-scale and multi-mode systems, 2) the nonlinear and hidden dependences embedded in the sensor data of various subsystems, and 3) the limitations of data collection. My research pivots around the development of analytical methods that leverage this large-scale, real-world operational data to enhance air transportation systems. In this talk, I will delve into our efforts to integrate physics and mechanism models with machine learning, optimization, and high-performance computing to fuel more informed modelling, prediction, and planning of air transport systems. I will showcase two key machine learning paradigms: 1) Supervised and unsupervised data-driven learning for uncovering insights and data mining in conventional airline and airport operations. 2) Dynamic programming and reinforcement learning for planning in the context of emerging air mobility technologies, such as designing route networks for urban drone delivery.