18th ERCOFTAC Da Vinci Competition 2022 - Interview with Da Vinci Finalists

Christian Lagemann 
(RWTH Aachen University, Germany)

Deep Recurrent Neural Networks for Optical Flow Learning in Particle-Image Velocimetry

• What is the topic of your Ph.D. thesis?

​In my PhD thesis, I investigated new ways to combine novel ideas of deep learning and experimental image-based fluid flow measurement techniques. Precisely, I introduced a learning based end-to-end optical flow algorithm called RAFT-PIV which is specifically tailored to Particle-Image Velocimetry experiments. In contrast to existing methods, which are mostly relying on cross-correlations, my algorithm effectively learns the prediction of dense velocity fields at a very high spatial resolution to estimate even tiniest flow structures.

• What does the Da Vinci competition mean to you?

To be counted as one of the participants in the Da Vinci competition is a great honor for me and strengthens my motivation to pursue my way in the academic landscape. It was a pleasure to meet these excellent PhDs from all over Europe and marked an outstanding possibility for deepening and extending research connections to various members of the ERCOFTAC and beyond.

• What did trigger your interest in STEM?

​My dad owns a very advanced workshop where I’ve spent a lot of time prototyping my own little projects when I was a kid. I guess this paved the way for my future work direction.

• What motivates you in your work?

​I’ve always loved to find new solutions to challenging problems and explanations to fundamental or also everyday challenges starting from small projects like finding the most efficient implementation of an algorithm to fundamental (mostly unsolved) research questions. And for me, this is actually what research is about – finding new or better suited explanations using state-of-the-art tools and algorithms.

• Where do you see yourself in five years?

​I will continue my way in fundamental research and hopefully will be allowed to supervise my own team of PhD candidates and graduate students.

• What advice would you give to new PhD students starting in fluid mechanics?

We should pay more attention to “older” literature instead of narrowly focusing on the last decade. Those researchers demonstrated an exceptionally high capability of explaining and deriving the underlying physics only based on comparably simple observations. To my experience, many of these earlier theories are incredibly useful to start a new research topic since they are much more intuitive.