- Title: AddBiomechanics – Lowering the Barriers to Musculoskeletal Modeling and Large-Scale Discoveries in Biomechanics
- Speakers: Keenon Werling, Stanford University
- Time: Wednesday, September 20, 2023 at 9:00 AM Pacific Time
Generating large-scale public datasets could unlock insights into human movement, neuromuscular diseases, and new treatments and interventions. However, processing movement data with detailed musculoskeletal models requires a substantial amount of time and expertise.To address this problem, Keenon Werling and colleagues at Stanford University developed AddBiomechanics, a free online tool with the mission of enhancing the impact of biomechanical motion capture efforts by helping labs process and share their data, unlocking powerful machine learning tools and improving replicability of results. AddBiomechanics automatically generates scaled models, joint angles, and torque trajectories that best fit your optical marker trajectories and force plate measurements with similar or better accuracy compared to expert-processed results. The uploaded data and processed results are automatically shared with the community, under a license that requires users of the data to cite your work.
In the first part of this webinar, Mr. Werling will present the AddBiomechanics web platform and show how it can fit into your lab’s workflow. He will outline the data sharing vision of the platform, describe how the tool works, discuss its evaluation of performance against human experts, and introduce the inverse dynamics features that have been recently added.
In the second half of the webinar, Mr. Werling will walk participants through a hands-on tutorial of uploading and processing motion capture data, provide best practices and troubleshooting tips when using AddBiomechanics, and provide an easy “getting started” kit for data sharing compliance.
Keenon Werling, Nicholas A. Bianco, Michael Raitor, Jon Stingel, Jennifer L. Hicks, Steven H. Collins, Scott L. Delp, C. Karen Liu. (2023) AddBiomechanics: Automating model scaling, inverse kinematics, and inverse dynamics from human motion data through sequential optimization. bioRxiv
Keenon Werling is a 3rd year PhD student in Computer Science and Biomechanics at Stanford University. He was diagnosed with Charcot Marie Tooth (type 2A) at age 8, and is working hard to get exoskeletons to work so that he can avoid a wheelchair as degeneration continues. His symptoms started out with mild foot-drop, and have progressed over the last 20 years to be complete paralysis below the knee, extreme weakness in his thighs, and some weakness in his hands.
Keenon did his undergrad at Stanford, class of 2016, and published two first-author papers in Chris Manning’s NLP Group. After undergrad, he founded Eloquent Labs with Gabor Angeli. Eloquent Labs built AI chatbots for customer support, grew to ten people, and was acquired by Square in 2019.
Now back at Stanford, Keenon has been working on curing movement disabilities with better powered exoskeletons. His work is supported by an NSF GRFP Award, and an NIH R01. That work started with a differentiable physics engine for simulating humans and machines, Nimble Physics, which forms the foundational technology that enables AddBiomechanics. With the data the community has shared on AddBiomechanics, Keenon is now working on using Machine Learning and custom wearable sensors to make better exoskeleton controllers. You can learn more about Keenon at keenon.github.io.