Seong Ho Yeon

I am a PhD candidate in the Biomechatronics group at MIT Media Lab advised by Prof. Hugh M. Herr. I am also a Graduate Research Fellow at K. Lisa Yang Center for Bionics at MIT. Prior to my PhD, I received M.S. in Media Arts and Sciences (MAS) from MIT in 2019, and I've received B.S. in Electrical Engineering from Georgia Institute of Technology and Korea Advanced Institute of Science and Technology (KAIST) in 2016. During my time at Georgia Tech, I received Roger P. Webb. ECE Senior Scholar Award.

My research interests lie in the intersection of bionics, robotics, and neural interfaces. The focus of my doctoral research is to enable versatile and seamless bionic applications by improving and expanding neural interfaces modalities and developing superior control architectures. The interdisciplinary nature of my PhD study has allowed me to encompass embedded system design, digital signal processing, modern robotics and optimization, and even neurosciences.

During my PhD study, I did internships at Google and the Robotics Lab at Hyundai Motor Company. Prior to my PhD, I interned in the Robotics Group at Naver Labs and the MX division at Samsung Electronics.

Email  /  Google Scholar  /  Media Lab Profile  /  linkedin

I view myself primarily as a system engineer. I like to a) understand basic building blocks, b) understand interfaces and interactions between these building blocks, and c) optimize networks of the building blocks with varied design criteria and limitations for creating synergized value and solving new problems.

Robotics recently became a symbolic pinnacle of system engineering. The field is a manifestation of multi-disciplinary system engineering. To me, it is intriguing that complex knowledge in dynamics, kinematics, non-linear optimization, path planning, and control theory gets consolidated into physical form based on the foundation of systematic integration with electrical and mechanical building blocks. I found that my background in varied system engineering put me in a unique position where I can understand and integrate low-level electrical system and operating system with advanced control theories and algorithms. Also, working with researchers in different disciplines help me develop a fluid mindset that transcends disciplines and learn system conventions across various fields of engineering. With that, I have made efforts to grow as a control system engineer with robotics emphasis.


PhD Research

Below is a list of different topics I studied during my PhD. The overarching theme of my PhD work is to advance control fidelity of bionic devices by leveraging modern knowledge in electronics, mechatronics, robotics, control theory, and optimization. Topics include: advancing an existing neural interface modality-Electromyography, pioneering a novel neural interface modality-Magnetomicrometery, improving scientific tools for neural interfaces research, and exploring closed-loop control approaches of bionic devices incorporating with real-time neural interface inputs.

Magnetomicrometry


Untethered Muscle Tracking Using Magnetomicrometry

Seong Ho Yeon#, Cameron Roy Taylor#, William H, Clark, Ellen G. Clarrissimeaux, Mary Kate O'Donnell, Thomas J. Roberts*, Hugh M. Herr*
Frontiers in Bioengineering and Biotechnology, Oct. 2022
MIT News / video / bioRxiv / dataset

We demonstrated real-time muscle tissue length tracking of the freely-moving turkeys executing various motor activities, including ramp ascent and descent, vertical ascent and descent, and free roaming movement using wireless Magnetomicrometry.

Clinical Viability of Magnetic Bead Implants in Muscle

Cameron Roy Taylor, William H, Clark, Ellen G. Clarrissimeaux, Seong Ho Yeon, Matthew J Carty, Stuart R Lipsitz, Thomas J. Roberts, Hugh M. Herr
Frontiers in Bioengineering and Biotechnology, Oct. 2022
bioRxiv

We validated the the clinical viability of magnetomicrometry (implants).

Magnetomicrometry

Cameron Roy Taylor, S. S. Srinivasan, Seong Ho Yeon, Mary Kate O'Donnell, Thomas J. Roberts, Hugh M. Herr
Science Robotics, vol.6, issue.57, Aug. 2021
MIT News / project page

We introduced a new sensing modality, magnetomicrometry, which uses the relative positions of implanted magnetic beads to wirelessly track tissue strains in real time. We demonstrated real time muscle length tracking in an in-situ turkey model via chronically implanted magnetic beads, while investigating accuracy, biocompatibility, and long-term implant stability.

On the first phase of the project, I validated and oversaw fabrication of the magnetic field sensing embedded system, and I was also heavily involved in setting up the real-time framework.

On the second phase of the project, I first redesigned the new magnetic field sensing electronics embedded system and improved power conditioning for the sensor. I also overhauled and reconstructed the whole magnetomicrometry computing framework. The current framework is able to interface multiple sensors with different communications means, and asynchronously acquire, process, and log data in real-time.

As the current framework becomes maturized and the two papers (in-situ and in-vivo validations) show promising results, I have been focusing on quantitative analysis and improvement of the current magnet(s) tracking algorithm (Levenberg-Marquardt based optimization).

Electromyography for Bionics


Acquisition of surface emg using flexible and low-profile electrodes for lower extremity neuroprosthetic control

Seong Ho Yeon, Tony Shu, Hyungeun Song, Tsung-Han Hsieh, Junqing Qiao, Emily A. Rogers, Samantha Gutierrez-Arango, Erica Israel, Lisa E. Freed, Hugh M. Herr
IEEE T-MRB, vol.6, issue.57, Jul. 2021

Flexible dry electrodes for emg acquisition within lower extremity prosthetic sockets

Seong Ho Yeon, Tony Shu, Emily A. Rogers, Hyungeun Song, Tsung-Han Hsieh, Lisa E. Freed, Hugh M. Herr
Biorob, 2020  (Oral Presentation, Nominated as a best student paper award candidate)
presentation / demo video / patent

This study presents an effective within-socket sEMG acquisition paradigm using a customized flexible and low-profile electrode designed to be compatible with prescribed liners and sockets. Quantitative analyses suggest comparable signal qualities between the custom and commercial electrodes while qualitative analyses suggest the feasibility of real-time sEMG data collection from load-bearing, ambulatory subjects with LE amputation.

Rejecting impulse artifacts from surface emg signals using real-time cumulative histogram filtering

Seong Ho Yeon, Hugh M. Herr
EMBC, 2021  (Oral Presentation)
presentation

This paper presents a cumulative histogram filtering (CHF) algorithm to filter impulsive artifacts within surface electromyograhy (sEMG) signal for time-domain signal feature extraction. The proposed CHF algorithm filters sEMG signals by extracting a continuous subset of amplitude-sorted values within a real-time window of measured samples using information about the probabilistic distribution of sEMG amplitude. For real-time deployment of the proposed CHF algorithm on an embedded computing platform, we also present an efficient, iterative implementation of the proposed algorithm.

Spatiotemporally Synchronized Surface EMG and Ultrasonography Measurement Using a Flexible and Low-Profile EMG Electrode

Seong Ho Yeon#, Hyungeun Song#, Hugh M. Herr
EMBC, 2021  (Oral Presentation)
presentation

We introduce a low-cost, noninvasive flexible electrode that provides high quality sEMG recording, while also enabling spatiotemporally synchronized ultrasonography recordings. Our results show no significant artifact in ultrasonography from both the phantom and TA fascicle strains due to the presence of the electrode, demonstrating the capability of spatiotemporally synchronized sEMG and ultrasonography recording.

Design of an Advanced sEMG processor for Wearable Robotics Applications

Seong Ho Yeon,
MIT Master Thesis, 2019

This thesis presents the design and evaluation of a surface electromyography(sEMG) acquisition platform specialized for wearable robotic applications. The thesis reasons and explains in detail every design decision and process among the system development and manufacturing.

This is the first version of the embedded EMG processor which we used for several bionics applications especially with the climbing ankle prosthesis project. In 2020, I designed the next version of the embedded EMG processor improving 1) system stability, 2) number of channels, 3) computing capabilities, 4) HW analog interface, and 5) digital I/O. While we did not write a manuscript on the new design of the EMG processor, my work now enables all bionics projects using real-time EMG inputs in the Biomechatronics group at MIT Media Lab.

I worked tp continuously improve embedded EMG system development for real-time bionics applications after I had developed the first version during my master's study. This new version has 1) doubled EMG input channels (16ch), 2) augmented safety with better isolation scheme (designed to be compliant to IEC 60601-1 standard), 3) more versatile digital I/O, 4) wider range of power input (7.5 ~ 60 V), and 5) more powerful computing capability (~x5). As this type of improvement and re-iteration are somewhat technical, I decided not to publish a stand-alone paper for the system development. Nonetheless, I wanted to share some technical details with people through this website. (Please refer to my master’s thesis for a discussion of the fundamentals.)

profile photo
profile photo

These are physical and conceptual block diagrams of the embedded EMG system. The system mainly consists of analog-front-end(AFE) block, digital-back-end block, and power conditioning and safety block.

profile photo

Here, I present a structure of high-SNR, high-input impedance, and high-dynamic-range DC-coupled AFE structure. Intrumentation amplifier buffers the signal with high-input impedance with small or no gains. Small gain buffering enables larger dynamic range of the input EMG signal. High-performance ADCs are then capable of acquiring un-amplified (but buffered) signal within a uV range.

profile photo

This is the power flow diagram of the embedded EMG system. Each of power conditioning IC takes specific purpose either maintaining efficiency, augmenting safety, or suppressing noise.

profile photo
profile photo

These two photos show example usage cases of the EMG system on bionics applications with different types of electrodes. In this case, the system was utilzied in a lower-extremity robotic prostheses system application. Green board is a second layer analog adapter for physical interfacing between the electrodes and the EMG system. The system provides real-time, processed EMG signal to mechatronic systems for closed-loop bionics applications.

The four projects in advancing EMG sensing modality were heavily motivated by lower-extremity prosthetic robotics applications. Efficacies of commercially-available EMG systems have been largely limited to able-bodied biomechanics research applications. In order to use EMG modality in prosthetic applications, we have to overcome different types of inter-disciplinary challenges. Flexible sensing electrode, impulse artifacts filtering, simultaneous ultrasound-EMG, and embedded EMG processing system are all actively used in exoskeleton and prosthetics applications research in our lab as of 2022. I personally considered these pulbications as research by-products. What really matters to me is that these advancements in EMG modality actually have enabled a lot of real-time bionics applications which would have been impossible otherwise.

Control Methods of Bionic Devices


Energy Efficiency and Performance Evaluation of an Exterior-Rotor Brushless DC Motor and Drive System across the Full Operating Range

Tsung-Han Hsieh, Seong Ho Yeon, Hugh M. Herr
Actuators, 2023

In recent years, exterior-rotor brushless DC motors have become increasingly popular in robotics applications due to their compact shape and high torque density. However, these motors were originally used for continuous operation in drones. For applications such as exoskeletons, prostheses, or legged robots, short bursts of high power are often required. This paper presents experimental data on the torque–speed relationship, efficiency, and thermal responses of one of the most widely used outrunner-type brushless motors across its full operating range, including high-power short-duration operation. The results of this study can inform the selection and design of actuators for a range of robotics applications, particularly those that require high power output for brief periods of time.

Design and Evaluation of a Quasi-Passive Variable Stiffness Prosthesis for Walking Speed Adaptation in People With Transtibial Amputation

Emily A Rogers, Seong Ho Yeon, Christian Landis, Hugh M. Herr
IEEE/ASME T-Mech, 2023

The biological ankle joint adjusts stiffness to adapt to changing walking speed, terrain, and load carriage. The most commonly used passive transtibial prostheses are unable to adjust device stiffness and therefore do not maximize potential energy storage and peak prosthesis power across speeds. We present a quasi-passive variable stiffness ankle-foot prosthesis with discrete stiffness adjustment from 352 - 479 Nm/radian, corresponding to the range of biological ankle quasi-stiffness exhibited during level ground walking at speeds from 0.75 - 1.5 m/s for a 77 kg person. We implement a novel parallel leaf spring mechanism that utilizes custom solenoid-driven linear actuators to constrain sliding of parallel leaf springs relative to a mechanical ground in order to control bending stiffness. The prosthesis is lower in mass than all existing variable stiffness prostheses, with a mass of 945 grams. We present initial results from a pilot study with one participant with unilateral transtibial amputation, demonstrating an increase in range of motion, peak prosthesis power, and energy storage and return, and a decrease in contralateral knee external adduction moment across a range of walking speeds. This variable stiffness ankle-foot prosthesis demonstrates the potential to improve biomechanics of walking through the design of a low-mass, quasi-passive prosthesis.

Modulation of Prosthetic Ankle Plantarflexion Through Direct Myoelectric Control of a Subject-Optimized Neuromuscular Model

Tony Shu, Christopher Shallal, Ethan Chun, Aashini Shah, Angel Bu, Daniel Levine, Seong Ho Yeon, Matthew Carney, Hyungeun Song, Tsung-Han Hsieh, Hugh M. Herr
IEEE RA-L, 2022

The ideal neuroprosthetic control paradigm minimizes the disparities in both dynamic performance and subjective embodiment between artificial and intact physiology. In this letter, we address the objective by introducing a cascaded optimization procedure known as the Activation Mapping paradigm that enables a subject with unilateral transtibial amputation to neurally modulate the torque developed during prosthetic ankle plantarflexion.

An ankle-foot prosthesis for rock climbing augmentation

Emily A Rogers, Matthew Carney, Seong Ho Yeon, Tyler R Clites, Dana Solav, Hugh M. Herr
IEEE T-NSRE, 2020

This research presents the design and preliminary evaluation of an electromyographically (EMG) controlled 2-degree-of-freedom (DOF) ankle-foot prosthesis designed to enhance rock climbing ability in persons with transtibial amputation. The results suggest that a lightweight, actuated, 2-DOF EMG-controlled robotic ankle-foot prosthesis can improve ankle and subtalar range of motion and climbing biomechanical function.

profile photo

Along with developing and advancing EMG sensing modality, I have been closely collaborating and leading multiple bionics applications (which use the custom EMG system). On most projects, I have led or participated in designing and implementing closed-loop control framework.

Please stay tuned for many upcoming projects. I'm also currently leading an interesting mechatronics project.


Selected Projects

I archived some interesting projects that I worked on outside of my PhD study. The list includes my class projects at MIT with emphasis on modern robotics and projects from my undergraduate years.

Robotics at MIT


Playing Piano with a Robotic Hand

MIT 6.843 Robotic Manipulation, Fall 2021, Taught by Dr. Russ Tedrake
Final Project
code / video / report

This project presents a robotic framework to play a mechanical piano with an anthropomorphic 23-DoF robotic arm-hand. The project was designed to incorporate essence of robotic manipulation including offline trajectory planning and dynamic model-based feedback controller. Whole implementation was constructed on the MIT DRAKE framework.

Swing up and balancing of inverted pendulum on a 2-D quadrotor

MIT 6.832 Underactuated Robotics, Spring 2019, Taught by Dr. Russ Tedrake
Final Project
code / report / presentation /
video1-single-swing-up / video2-double-swing-up /
video3-single-dynamics / video4-double-dynamics

We present a hybrid controller for swing-up and balancing of an inverted single and double pendulum on a flying 2-D quadrotor platform based on dynamics analysis. Lyapunov analysis on synthesized LQR controller and insights about 10-DoF dynamics of the quadrotor played as key building blocks for the project.

Undergraduate Research


A Portable and Wireless Transcutaneous Electrical Nerve Stimulation System to Generate a Pressure Sensation on the Foot

Seong Ho Yeon, B. Saravanabhavan, A. D. Filippo, Hangue Park*, Stephen P. DeWeerth*
EMBC, 2016  (Research Poster Presentation)
paper / poster
Georgia Tech UROP

The development of a diabetic peripheral neuropathy (DPN) ultimately leads to amputation due to the abnormal pressure distribution caused by loss of sensation. This work presents a portable and wireless transcutaneous electrical nerve stimulation (TENS) system prototype to emulate foot pressure sensation for diabetic neuropathy patients in order to recover the abnormal pressure distribution and prevent complications of diabetic peripheral neuropathy.

All electrical and real-time ECG, respiration, airflow, and skin conductance monitoring system

Jonghwa Lee, Seong Ho Yeon, SeongHwan Cho
ISOCC, 2015  (Oral Presentation)
presentation
KAIST URP

We present a low-cost and real-time multiple physiological parameter monitoring system that measures ECG, respiration, airflow and skin conductance by using an all-electrical method. Skin conductance is chosen to evaluate the activity within the sympathetic axis of the autonomic nervous system. Bio-impedance (BI) and resistance measurement technique using DC servo loop are employed to achieve allelectrical measurement. The system was implemented with off-the-shelf components and the total cost of the components is less than a dollar. Measurement results show that ECG, respiration, airflow and skin conductance can be clearly measured simultaneously without interference.

These undergraduate projects may seem pre-matured. Through these projects, I hope you could see the process of my development and growth as an engineer.


Work Experiences

Throughout my undergraduate and graduate studies, I've had three internship opportunities. Each of internship provided unique insights and experiences.

...

Emerging Sensor System & Mechatronics Hardware Engineering Internship

TechEng Sensors Team,
Pixel Business Unit,
Google Inc.
May. 2023 ~ Aug. 2023

During this internship, I collaborated with PhD hardware engineers to explore and develop novel sensor systems for next generation Pixel (+ecosystem) devices. The scope of work includes control system design, embedded system design, mechatronics development, and digital signal processing.

Control of Dynamic and Articulated Robots

Robot Platform Team, Robotics Lab
Hyundai Motor Company
May. 2020 ~ Aug. 2020

During the internship, I conducted research on developing control scheme of articulated dynamic robots. The first project involved exploring and demonstrating closed-loop control simulations for dynamic and mobile robots, including a 6-DoF robot-arm manipulator and a mobile quadruped robot. The second project focused on researching the control system and algorithm of the 6-DoF medical lower-limb exoskeleton, H-MEX2, aimed at assisting and restoring the ambulatory activity of a paraplegic person. I was mentored by Dr. Dong Jin Hyun during the internship.

Design and Implementation of Modern Mobile Robot Architecture

Robotics Group
Naver Labs
Jan. 2017 ~ Jul. 2017

During this internship, we designed and implemented a mobile robot (hardware and software) framework from scratch with a team of interns. This project led to two publications during the internship and has been transformed to a Senior Capstone Design Project Course of KAIST MechE department. I was mentored by Dr. Sangok Seok during the internship.

Applying Asynchronous Deep Classification Networks and Gaming Reinforcement Learning-Based Motion Planners to Mobile Robots

Seong Ho Yeon#, Gilhyun Ryou#, Youngwoo Sim#, Sangok Seok
ICRA, 2018
video

System design for autonomous table tennis ball collecting robot

Seong Ho Yeon#, Dayeon Kim#, Gilhyun Ryou#, Youngwoo Sim#
ICCAS, 2017

...

Embedded Hardware Engineering Internship

Innovative Solution Part, Flagship R&D Team,
Mobile Communication Business,
Samsung Electronics
Jun. 2016 ~ Aug. 2016

I participated in this internship at the team developing a mobile sensor solution for next-gen flagship smartphone(Galaxy) of Samsung. During the internship, I mostly learned about rigorous industrial process to design and develop commercial devices for mass-production. As a small contribution within the team, I proposed the concept of next-generation optical heart rate monitor (HRM) sensor with auto-calibration feature to compensate individual differences of people to achieve higher robustness of the system. I also participated in ideation and evaluation of varied next-generation disruptive sensor solutions for mobile products such as environment sensors or bio-sensors.


Exhibitions

Adventures in Bionics #FutureTechFriday

Adventures in Bionics-#FutureTechFriday, Filmed at MIT Media Lab, July 27. 2022
From SNS of the Mr. Robert Downey Jr.
Live demonstration of Magnetomicrometry during a launch event of K. Lisa Yang Center for bionics at MIT. Mr. Robert Downey Jr. has been one of executive advisory board members. It was such an honor to have the opportunity to demonstrate our recent work to iron man. The posted video is a short version. Please checkout the full video at the link.

Engineering for Sports

Engineering Stories, Engineering Design Workshop, Museum of Science, Boston, Permanent Exhibition of Museum of Science, Dec. 2021
Presenting engineering stories with neurally controlled rock climbing ankle prosthesis system.

Using A.I. to build a better human

Using A.I. to build a better human, The Age of A.I. Season 1 Episode 3, Youtube Original, Dec. 2019
Demonstrating neurally controlled rock climbing ankle prosthesis system.

This MIT Engineer Built His Own Bionic Leg

This MIT Engineer Built His Own Bionic Leg, Bloomberg Documentary, Dec. 2019
Demonstrating neurally controlled ankle prosthesis system. (I did not appear in the video)

How we’ll become cyborgs and extend human potential

Talk by Dr. Hugh M. Herr, How we’ll become cyborgs and extend human potential, TED 2018 Conference, Apr. 2019
Demonstrating neurally controlled ankle prosthesis systems (dynamic powered ankle and quasi-passiverock climbing ankle).
I was introduced as a member of "Team Cyborg" at MIT.



Website Design Credits: Jon Barron
Augmented with Bootstrap for mobile friendly website.