For a one-page description of my professional career, please see:
The common thread to my career in robotics, neuroscience, and software development is that I work as a “research engineer” who writes software to solve a difficult, large-scale, or mathematical task. Developing hypothesis, constructing models, fitting data, analyzing patterns, optimizing – it is always the same old math applied to different contexts.
My project portfolio contains a detail-oriented but conversational description of some of my past projects:
Presentations and Slides
After leaving the academic sphere, I have not given many public talks or had the opportunity to attend as many as I would otherwise enjoy. I wish I had recorded more of them than just these few!
I gave a fun “artsy” talk at Donut.js a few years ago, titled How Sunflowers Lead to Another Musical Scale. As a (very) amateur musician, this meagre contribution to musical theory may simply to have been to define the most dissonant dyadic chord possible.
Fiber Reinforced Polymers in Robotics
I’m not really an expert, but I have worked enough with fiberglass, foam, and honeycomb construction to give a seminar on its uses in robotics. There are some slides from 2012:
And if you are very patient and want to hear the whole thing:
Auditory Neural Models
I gave a talk at the Oregon Health & Science University on some research I was working on in 2014; unfortunately it was not recorded, so you’ll have to just read the slides.
I’ve given a few minor seminars on functional programming – more on the practical aspects of it and languages like Clojure than anything to do with Category Theory, monads, monoids, and all that.
Still, as I get older, I always feel like there is more timeless knowledge to be learned about programming if one asks the mathematicians rather than engineers.
Art is never finished, only abandoned. – Leonardo da Vinci
A genius far ahead of his time, Leonardo’s words ring particularly true for my thesis project, as I am sure they do for anyone else who has completed an art project of that magnitude. A Ph.D. is inherently artistic because it is an opportunity to be driven to strange places by intense curiosity and passion, and the whole process drives you a little crazy.
My thesis was pretty strange. I spent 3 years studying compliant actuators, and then in the last 3 months, I tried to construct a one-legged kangaroo robot that used the compliant actuators:
If you are curious about it, I would recommend starting by skimming through my JIMI Robot Ph.D. Thesis Defense Slides, or by watching me rush through the thesis presentation, and questions, in a mere 35 minutes:
My thesis title was tongue-in-cheek with regard to academic writing style – years later, it now makes me both groan and giggle because it is so long that even I cannot remember it all. If you want to dig into the thesis in all its beautiful imperfections, here it is:
- JIMI: A Hopping Monopod Robot Incorporating Nonlinear Series Elastic Actuators, Fiber-Reinforced Polymer Construction, and a Concurrent Asynchronous Dataflow-based Centroidal Momentum Balance Controller
My masters thesis at Nagoya University mostly concerned with studying via simulation the intrinsic stability of passive dynamic robots. To challenge myself, I gave the final presentation in Japanese, but despite my best efforts, the English translation is doubtlessly more grammatically correct.
- On the Gait Robustness of Passive Dynamic Robots, and a Novel Variable Stiffness Series Elastic Actuator
- Slides for the above thesis (Japanese)
- Slides for the above thesis (English)
I would like to deeply thank the people of Japan for their generosity and exchange scholarship programs. Without the MEXT scholarship having paid for my tuition and living expenses, I would never have seen the world, and my life would have been much the poorer. Hontou ni arigato goziamasita!
Open Source Software
I wrote the first and most of the second version of NEMS, which is open source software for modeling the functional properties of auditory neurons. Prof. Stephen David and I used the software extensively in my lone publication in neuroscience. It is still being used to this day in the LBHB lab, and if the code doesn’t suffer too much “academic code drift”, my hope it that it will help other auditory neuroscientists do more sophisticated neural functional modeling as well.
US 8821338: Elastic Rotary Actuator
ITALY 0001407702: Attuatore Rotante Elastico con Meccanismo Ipocicloida
Basically, both are essentially the same patent, for this machine:
Thorson, Ivar. Liénard, Jean. David, Stephen V. The Essential Complexity of Auditory Receptive Fields PLOS Computational Biology, Dec 18, 2015.
Thorson, Ivar. Caldwell, Darwin G. “A Nonlinear Series Elastic Actuator for Highly Dynamic Motions”. IROS 2011. DOI: 10.1109/IROS.2011.6048581.
Vanderborght, Bram. Tsagarakis, Nikos. Van Ham, Ronald. Thorson, Ivar. Caldwell, Darwin G. “MACCEPA 2.0: Compliant actuator used for energy efficient hopping robot Chobino1D”. Autonomous Robots 31(1)55-65.
Thorson, Ivar. Svinin, Mikhail. Hosoe, Shigeuki. Asano, Fumihiko. Taji, Kouichi. “Quantifying Gait Robustness of Passive Dynamic Robots.” May 2007.
Thorson, Ivar. Svinin, Mikhail. Hosoe, Shigeuki. Asano, Fumihiko. Taji, Kouichi. “Design Considerations for a Variable Stiffness Actuator in a Robot that Walks and Runs.” May 2007.
I also have some now “declassified” videos of a project I worked on at ATR a decade ago, which was to make a hybrid pneumatic/electric exoskeleton that could provide 0 to 100% augmentation of strength or balance. The idea was for the pneumatic cylinders to provide the large, low-frequency forces and torques, and the high-frequency torques to be handled by the electric actuators.
Prof. Hyon also did an impromptu filming of a rather wild-haired me demonstrating the full-body posture control system that I cooked up when I was supposed to be working on other things. I loved that little white Japanese eeePC, because I could carry it in my cargo pants’ pockets and walk around controlling the robots with the joystick, while staying safely out of reach of the very powerful humanoid hydraulic robot. Partly I was afraid of being crushed, and partly just the fact that, at 3000PSI, even a pinhole leak will easily inject oil through your skin.
We tried to do make some trajectories to get the robot to mimic baseball batting, but the work never came to anything as far as I know. For more information on these projects, please see the work of Prof. Sang Ho Hyon.
Most of the robots that I have built outside of work in recent years are robots intended for tutorials when teaching a class of students, or as gifts if I have some extra spare parts (easy to find) and some spare time (hard to find!).
Going much further back in time, I made this in 1997 using the good old Motorola 68HC11, an assembler, some carbon fiber sheets, a table saw, and lots of drill bits…