CDC’18 in Miami Beach

I attended the 2018 Conference on Decision and Control in Miami Beach, Florida. A welcome change of weather from the looming winter in Madison! The photo above was taken from the terrace of the conference venue.

At the conference, I took part in the Workshop on the Intersections of Machine Learning and Parameter Estimation in Control organized by Travis Gibson, Joseph Gaudio, and Anuradha Annaswamy of MIT. This general field at the intersection of machine learning, optimization, and control seems to be gaining momentum. Indeed, there was another workshop on very similar topics that was held concurrently and it was a shame I couldn’t attend both!

I’ll point out two highlights of the conference. Besides attending sessions on robust control, distributed optimization, decentralized control, and machine learning, I also strayed outside my comfort zone and attended a tutorial session on “Control-Theoretic Methods for Biological Networks”. A growing group of controls researchers have taken interest in biological systems, and I have to say I was both impressed at the progress in this field and humbled at how much there is left to do. Biological systems are incredibly complex and it seems the only way to advance the ball (and to be taken seriously) in this area is to work at the problem from both sides: performing biological experiments in a wet-lab environment and doing some serious mathematical modeling and analysis. I thought Hana El-Samad (UCSF) and Eduardo Sontag (Northeastern) both gave very insightful presentations.

The second highlight was a special session commemorating the life and contributions of Bruce Francis to the field of robust control. Bruce passed away in March of this year and it was nice to see so many people (standing room only) come to celebrate his life and accomplishments at CDC. Bruce’s (arguably) biggest contribution was in developing a state-space solution for the H-infinity control problem. The famous “DGKF” paper is among the most highly-cited papers in the field. The other three co-authors (Doyle, Glover, and Khargonekar) were all in attendance and shared some heartfelt words. The first photo on the right was taken during the CDC special session; John Doyle was recounting some of the initial (mixed) reactions to the DGKF paper, including some choice quotes (see the slide!). The first (and last) time I met Bruce Francis was at the Caltech CDS20 workshop five years ago. Incidentally, all four authors of DGKF were present at that event (see second photo on the right — authors are in the DGKF order!). Although Bruce and I had just met, we bonded over the fact that we are both Canadian, and we both enjoy playing Frisbee! Bruce was also a Professor at the University of Toronto, which is where I did my undergrad. Bruce was soft-spoken, kind, and humble — sentiments echoed by many of his colleagues at CDC this year. He will be missed.

This year’s Bode lecture was delivered by Mark Spong of UT Dallas. The topic was robotics, an area that is not really in my wheelhouse, but I thought Mark did a wonderful job of explaining some basic tools and challenges from the field. What amazed me from the presentation was how simple passivity-based control can be used to create very natural-looking movement patterns that are robust to disturbances. Mark showed examples of a self-righting inverted pendulum and various walking robots, all of which were designed using energy conservation principles. If you’re looking for a well-delivered introductory lecture on robotics, I highly recommend this one! The lecture was recorded and should be posted online here soon.