inControl

In this episode, our guest is Anuradha Annaswamy. Anu is the Director of the Active-Adaptive Control Laboratory and Senior Research Scientist at the Massachusetts Institute of Technology in the Deparment of Mechanical Engineering.  We delve into adaptive control and its exciting history, ranging from the Brave Era to the audacious X15 tests and to modern intersections with Reinforcement Learning.

Outline
02:15 - Anu's background
05:20 - What is adaptation?
08:30 - The Brave Era
15:17 - The X15 accident 
23:16 - Exploration vs exploitation
28:35 - Beyond linearity and time invariance
45:05 - Adaptive control vs Reinforcement Learning
52:12 - The future of adaptive control
54:34 - Outro

Episode links
Anu's lab:  http://aaclab.mit.edu/
NCCR Symposium: https://tinyurl.com/bdz84p4c
Book - Stable adaptive systems: https://tinyurl.com/mw4saame
X-15 Flight 3-65-97: https://tinyurl.com/2kbe7nsy
Paper - Adaptive Control and the NASA X-15-3 Flight Revisited: https://tinyurl.com/2p83k7ez
Paper - A historical perspective of adaptive control and learning: https://tinyurl.com/yck89rcd
Paper -Adaptive Control and Intersections with Reinforcement Learning: https://tinyurl.com/yc27rsyd
KYP Lemma: https://tinyurl.com/mkf35jjt
Persistence of excitation: https://tinyurl.com/bpfwp9n9
Dual control: https://tinyurl.com/ywduzm5x
Paper - Robust adaptive control in the presence of bounded disturbances:  https://tinyurl.com/4pztx23z
Paper -  Reinforcement learning is direct adaptive optimal control https://tinyurl.com/appnjzyn
MRAC: https://tinyurl.com/bdzzphju
Self Tuning Control: https://tinyurl.com/3mjs3skm

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

In this episode, our guest is Jean-Jacques Slotine, Professor of Mechanical Engineering and Information Sciences as well as Brain and Cognitive Sciences, Director of the Nonlinear Systems Laboratory at the Massachusetts Institute of Technology, and Distinguished Faculty at Google AI.  We explore and connect a wide range of ideas from nonlinear and adaptive control to robotics, neuroscience, complex networks, optimization and machine learning.

Outline
00:00 - Intro
00:50 - Jean-Jacques' early life
06:17 - Why control?
09:45 - Sliding control and adaptive nonlinear control
18:47 - Neural networks
23:15 - First ventures in neuroscience
28:27 - Contraction theory and applications
48:26 - Synchronization
51:10 - Complex networks
57:59 - Optimization and machine learning
1:08:17 -  Advice to future students and outro

Episode links
NCCR Symposium: https://tinyurl.com/bdz84p4c
Sliding mode control: https://tinyurl.com/2s45ra4m
Applied nonlinear control: https://tinyurl.com/4wmbt4bw
On the Adaptive Control of Robot Manipulators: https://tinyurl.com/b7jcpkzw
Gaussian Networks for Direct Adaptive Control: https://tinyurl.com/22zb7pkx
The intermediate cerebellum may function as a wave-variable processor: https://tinyurl.com/2c34ytep
On contraction analysis for nonlinear systems: https://tinyurl.com/5cw4z9j8
Kalman conjecture: https://tinyurl.com/2pfjsbke
I. Prigogine: https://tinyurl.com/5ct8yssb
RNNs of RNNs: https://tinyurl.com/3mpt7fec
How Synchronization Protects from Noise: https://tinyurl.com/2p82erwp
Controllability of complex networks: https://tinyurl.com/24w7hdae
B. Anderson: https://tinyurl.com/e9pkyxdx
Online lectures on nonlinear control: https://tinyurl.com/525cnru4

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

In this episode, we delve into the extraordinary life of Norbert Wiener, the founding father of cybernetics - the science “control and communication in the animal and the machine”.

Outline
00:00 - Intro
02:06 - The early years of Norbert
09:00 - Europe and WWI
15:50 - MIT days
19:30 - Norbert’s marriage
22:39 - Generalised harmonic analysis
28:18 - The interactions with Hopf and Paley
31:14 - Bush and the analog computer program
35:55 - WWII, Bigelow and prediction theory
40:41 - Rosenbleuth and teleological machines
47:56 - Mexico and Norbert’s biological investigations
51:25 - Cybernetics
1:00:16 - The human behind Norbert Wiener
1:01:53 - Outro
 
Episode links
Things named after Wiener: https://tinyurl.com/mt37xn93
Autobiography: https://tinyurl.com/2umws9nd
Biography: https://tinyurl.com/nhawc9az
Wiener filter: https://tinyurl.com/n9u5ukxe
Paley-Wiener theorem: https://tinyurl.com/mr3z3f89
Wiener-Kinchin theorem: https://tinyurl.com/3mxm54ac
Vannevar Bush: https://tinyurl.com/y6s7kz6t
Julian Bigelow: https://tinyurl.com/28m4a6as
Behavior, Purpose and Teleology: https://tinyurl.com/3ut2afjz
Arturo Rosenblueth: https://tinyurl.com/57wp67vh
Cybernetics:  https://tinyurl.com/5e3tnn6e
Out of control:  https://tinyurl.com/3rnhn3xh
A scientist rebels: https://tinyurl.com/5f2d3urc
Moral and technical consequences of automation: https://tinyurl.com/72tvzuxy

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

In this episode, our guest is Sean Meyn, Professor and Robert C. Pittman Eminent Scholar Chair in the Department of Electrical and Computer Engineering at the University of Florida. The episode features Sean’s adventures in the areas of Markov chains, networks and Reinforcement Learning (RL) as well as anecdotes and trivia about beekeeping and jazz.

Outline
00:00 - Intro
00:22 - Sean’s early steps
03:53 - Markov chains
08:45 - Networks
18:26 - Stochastic approximation
25:00 - Reinforcement Learning
38:57 - The intersection of Reinforcement Learning and  Control
42:37 - Favourite theorem
44:05 - Beekeeping and jazz
48:47 - Outro

Episode links
Sean’s website: https://meyn.ece.ufl.edu/
Sean’s books: shorturl.at/CFGRY (and T. Sargent's review: shorturl.at/hlGNR)
G. Zames: shorturl.at/JPRWX (see also: shorturl.at/chiw5)
State space model: shorturl.at/hST07
The life and work of A.A. Markov: shorturl.at/qsv35
Fluid model: shorturl.at/HKN56
M/M/1 queue: shorturl.at/dQW36
Borkar-Meyn theorem: shorturl.at/eSTV4
NCCR Automation Symposia: shorturl.at/csv03 (see also shorturl.at/ekpZ3)
V. Konda’s PhD Thesis: shorturl.at/bdrv7

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

In this episode, our guest is Alessandro Chiuso. Alessandro is a Professor in the Department of Information Engineering at the University of Padova. The episode covers several topics, including Alessandro’s research trajectory, his work in system identification and vision, and his passion for skiing. Check out Alessandro’s website here: http://automatica.dei.unipd.it/people/chiuso.html

Outline
00:00 - Intro 
01:51 - Research trajectory
03:52 - Influential figures 
08:20 - System identification
17:07 - Regularized system identification 
23:30 - Vision
28:40 - Data-driven methods
30:32 - Future of system identification
33:40 - Question from the audience
34:19 - Advice to future students
35:50 - Skiing at a semi-professional level

Episode links
Giorgio Picci's website: http://www.dei.unipd.it/~picci/
Stefano Soatto's website: http://web.cs.ucla.edu/~soatto/
ERNSI: https://people.kth.se/~bo/ERNSI/
System identification: https://en.wikipedia.org/wiki/System_identification
Regularized system identification:  https://tinyurl.com/yc7b7myt
Origin of “regularization”: https://tinyurl.com/y4jmk75f
Hirotugu Akaike: https://en.wikipedia.org/wiki/Hirotugu_Akaike
Structure from motion: https://tinyurl.com/35canfnx
Dynamic textures: https://tinyurl.com/28bdwhwm
Skiing:  https://tinyurl.com/2p8xzau6

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

In this episode, our guest is Ben Recht. Ben is a Professor in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley.  We discuss several topics, including his research trajectory, Ben's tour of reinforcement learning, and his passion for music, among others. Check out Ben's website here: http://people.eecs.berkeley.edu/~brecht/

Outline
00:00 - Intro
01:01 - Ben predicts the birth of "inControl"
02:40 - Personal research trajectory
06:55 - How and why did you dive into control theory?
08:43 - Influential figures who shaped Ben's research
13:50 -  The "argmin" blog &  myth busting
27:43 - Ben's tour of reinforcement learning
45:18 - Future challenges for control
52:06 - Biological origin of learning
58:24 - "This or that" game
1:02:54 - Questions from the audience
1:14:51 - What would you do if you were a student today?
1:17:00 - Ben's band: "the fun years"

Episode links
Ben's website: http://people.eecs.berkeley.edu/~brecht/
argmin: http://www.argmin.net/
the fun years: http://thefunyears.com/
A tour of reinforcement learning: https://arxiv.org/abs/1806.09460
Patterns, predictions and actions: http://mlstory.org/
System level synthesis: https://arxiv.org/abs/1904.01634
Aizerman's conjecture: https://en.wikipedia.org/wiki/Aizerman%27s_conjecture

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

This episode features an interview with Florian Dörfler, who is an Associate Professor at the Automatic Control Laboratory at ETH Zürich, Switzerland.  We discuss several topics, including his personal research trajectory, the influence of machine learning on control, future challenges in control theory, among others. Check out Florian's website here: http://people.ee.ethz.ch/~floriand/

Outline

00:00 - Intro 
01:03 - Personal research trajectory
05:57 - Influence of machine learning on control
07:52 - Why doing research in control?
09:51 - What would you change in control? 
11:36 - Where is the field heading?
14:20 - Favourite theorem in control theory
16:20 - Vision: what would you like to achieve?
17:03 - Influential figures
19:17 - Sociology and control
21:23 - What would you do if you were a student today?

Episode links

Florian's website: http://people.ee.ethz.ch/~floriand/
Gerschgorin theorem: https://en.wikipedia.org/wiki/Gershgorin_circle_theorem
Synchronization paper: https://www.pnas.org/doi/abs/10.1073/pnas.1212134110
Hamming - "A stroke of genius": https://www.mccurley.org/advice/hamming_advice.html 

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.

This episode breaks the ice with a bit of the pre-history of control theory. We discuss three iconic ancestors of the science of feedback, including water clocks developed by Ktesibios, the earliest known thermostat,  and governors, a class of mechanical devices,  which,  without exaggeration, have enabled the first industrial Revolution in Britain.

Outline
00:00 -Intro 
01:32 - Ktesibios
06:15 - Cornelis Drebble
11:55 - Governors

Episode links

• O. Mayr - The origins of feedback control
• K. Kelly - Out of Control

Ktesibios

• https://en.wikipedia.org/wiki/Ctesibius

Drebble

• https://en.wikipedia.org/wiki/Cornelis_Drebbel
• https://nautil.us/issue/12/feedback/the-vulgar-mechanic-and-his-magical-oven
• https://sites.google.com/site/ukdrebbel/

Governors

• J.C. Maxwell, “On Governors,”Proc. of the Royal Society of London, vol. 16, pp. 270-283, 1868.
• S.Bennett- A History of Control Engineering 1800-1930
• Special issue on control education - The United Kingdom, by M.C. Smith, IEEE Control Systems Magazine, pp. 51-56, April 1996 (check also here).  

Support the show

Podcast info
Podcast website: https://www.incontrolpodcast.com/
Apple Podcasts: https://tinyurl.com/5n84j85j
Spotify: https://tinyurl.com/4rwztj3c
RSS: https://tinyurl.com/yc2fcv4y
Youtube: https://tinyurl.com/bdbvhsj6
Facebook: https://tinyurl.com/3z24yr43
Twitter: https://twitter.com/IncontrolP
Instagram: https://tinyurl.com/35cu4kr4

Acknowledgments and sponsors
This episode was supported by the National Centre of Competence in Research on «Dependable, ubiquitous automation» and the IFAC Activity fund. The podcast benefits from the help of an incredibly talented and passionate team. Special thanks to L. Seward, E. Cahard, F. Banis, F. Dörfler, J. Lygeros, ETH studio and mirrorlake . Music was composed by A New Element.