My research interests include modeling, controls, and optimization applied to human movement. I’ve worked on
both quantifying stability and uncovering underlying optimality criteria in the sit-to-stand motion. I aim to increase safety in wearable robotic systems, such as exoskeletons and prostheses, during dynamic tasks.
I grew up in Bothell, WA and graduated from the University of Portland with a B.S. in Civil Engineering in May 2016.
I am a National Science Foundation Graduate Research Fellow.
Inverse Optimal Control
Given a set of observations from a dynamic human task, inverse optimal control (IOC) assumes that the data are the result of an optimal control problem and aims to uncover the cost function minimized by this data. Once this optimality criteria is established, the cost can be used to predict how a human may respond to a perturbation during motion. Knowledge of how a human reponds to disturbances can be useful in designing controllers for wearable robotics.
In my IOC work, I use experimental observations from 11 subjects during the sit-to-stand motion to compare cost functions between subjects, connect the cost functions to physical quantities, and investigate how well IOC can predict the response to a cable-pull perturbation.
Fall risk can be mitigated by identifying those who are unstable and providing targeted physical therapy. However, no existing method of quantifying stability in human motion has been shown to reliably predict fallers. We use dynamic modeling and reachability to create Stability Basins, which characterize the set of perturbations one can withstand while completing a dynamic task.
To investigate whether our method correctly estimated the stable region, we conducted an experiment where subjects were perturbed by motor-driven cable pulls during the sit-to-stand motion. We formed and validated individualized Stability Basins for 11 subjects, confirming that our analysis accurately predicted the instances where the subject stepped or sat back down in response to perturbation.
The figure above shows all successful sit-to-stand trials contained within the Stability Basin, and an example of a perturbed trial where failure was predicted before the step occurred.
Evolutionary Hypotheses using Engineering Tools
I analyze animal biomechanics with Prof. Talia Moore to uncover explanations for patterns in evolution and ecology. Often, engineering and robotics can help us test hypotheses and learn more about animal movement.
The figure above describes a kinematic analysis of Micrurus coral snakes to quantify non-locomotory thrashing behavior. (A) shows a frame at the conclusion of a bout of thrashing, (B) describes the method for approximating curvature at the point of interest, (C) shows the curvature vector for each sampled point along the snake centerline, and (D) shows these curvature magnitudes plotted to the left and right of the snake.
Automated Rehabilitation Assessment
Physical therapists qualitatively assess patients' performance on a leg press machine as they recover from anterior cruciate ligament (ACL) reconstruction surgery. In addition to a clinician's expertise, measuring forces during this activity could provide a quantitative assessment of rehabilitative progress. This project utilized a stereo camera coupled with publicly available pose estimation software to accurately estimate the forces produced by patients.
In the figure above, (a) depicts the approximate stereo camera view as well as the estimated 2D positions of the participant's joints, (b, c, d) show the estimated shank, thigh, and trunk positions at the beginning, middle, and end of a leg press repetition, and (e) plots the estimated force on the foot plate from video data alone.
S. M. Danforth, M. Kholer, D. Bruder, A. R. Davis Rabosky, R. Vasudevan, T. Y. Moore, "Emulating duration and curvature of coral snake anti-predator thrashing behaviors using a soft-robotic platform," IEEE International Conference on Robotics and Automation, In Review, 2019.
P. D. Holmes, S. M. Danforth, X.-Y. Fu, T. Y. Moore, and R. Vasudevan, "Characterizing the limits of human stability during motion: perturbative experiment validates a model-based approach for the Sit-to-Stand task," Royal Society Open Science, In Review, 2019. [arXiv preprint]
T. Y. Moore, S. M. Danforth, J. G. Larson, and A. R. Davis Rabowski, "A kinematic analysis of Micrurus coral snakes reveals unexpected variation in stereotyped anti-predator displays within a mimicry system," 2019. [bioRxiv preprint]
C. H. Kim, S. M. Danforth, P. D. Holmes, D. R. Raz, D. Yao, A. Bedi, and R. Vasudevan, "Automated camera-based estimation of rehabilitation criteria following ACL reconstruction," 2018. [arXiv preprint]
S. M. Danforth, J. T. Martz, A. H. Root, E. B. Flynn, and D. Y. Harvey, "Multi-source sensing and analysis for machine-array condition monitoring," in Proceedings of the SEM XXXV International Modal Analysis Conference, Garden Grove, CA, January 2017. [url]
T. A. Doughty, L. J. Cassidy, S. M. Danforth, and N. Pendowski, "Varied system geometry and noise implementation applied to nonlinear model tracking," in Proceedings of the ASME 2016 International Mechanical Engineering Congress and Exposition, Phoenix, AZ, November 2016. [url]
T. A. Doughty, L. J. Cassidy, and S. M. Danforth, "Implementing noise, multi-frequency stimulus, and realtime analysis to nonlinear model tracking," in Proceedings of the SEM XIII International Congress, Orlando, FL, June 2016. [url]
S. M. Danforth, J. G. Larson, A. R. Davis Rabowski, and T. Y. Moore, "A Kinematic Analysis of Micrurus Coral Snake Thrash Duration and Curvature Enables Quantitative Characterization of Non-Locomotory Behavioral Motion," Society for Integrative and Comparative Biology, In Review, 2019.
S. M. Danforth, P. D. Holmes, and R. Vasudevan, "Inverse optimal control with sit-to-stand data," presentation in Dynamic Walking, Canmore, Alberta, Canada, June 2019.
P. D. Holmes, X.-Y. Fu, S. M. Danforth, T. Y. Moore, and R. Vasudevan, "N-step reachability to characterize human mediolateral stability during perturbed walking," poster presentation in Dynamic Walking, Canmore, Alberta, Canada, June 2019.
P. D. Holmes, S. M. Danforth, T. Y. Moore, and R. Vasudevan, "Perturbative sit-to-stand experiment for validating a quantitative method for stability estimation," presentation in World Congress of Biomechanics, Dublin, Ireland, June 2018.
P. D. Holmes, S. M. Danforth, T. Y. Moore, and R. Vasudevan, "Perturbative sit-to-stand experiment for validating a quantitative method for stability estimation," presentation in Dynamic Walking, Pensacola, FL, May 2018.
P. D. Holmes, S. M. Danforth, T. Y. Moore, X. Y. Fu, and R. Vasudevan, "Humans minimize error in task-relevant dimensions during sit-to-stand," presentation in Dynamic Walking, Mariehamn, Finland, June 2017.
Honors and Awards
Accepted into Rising Stars in Mechanical Engineering workshop, Stanford University
Graduate Research Fellowship Program Recipient, National Science Foundation
Education Foundation Scholarship Recipient, American Institute of Steel Construction
Tau Beta Pi Scholarship Recipient
A highlight of my time as a Ph.D. student has been creating figures and videos to ensure my work is presented in a straightforward manner. Entertaining videos can also make STEM research more accessible to a non-technical community. I am interested in making science communication a central part of my career after finishing my degree.
Here's the latest video I wrote and edited, starring my labmate Patrick: