Patrick Franks

I'm recruiting PhD students for Fall 2023; If you are interested in either advanced mechatronic design or machine learning-based control of assistive devices, please reach out! Examples of past projects on my lab site: shepherdlab.sites.northeastern.edu

@inseungkang @Michael_J_0 @KarlZelik @doctorBobG @g_m_bryan 400 hours well spent 😅

@Michael_J_0 @KarlZelik @doctorBobG Yes, 'learning' how to effectively use an exo is important. I remember having a chat with @p_franks_ @g_m_bryan asking about the number of hours they've walked in their exos during grad school. But, it is worth noting that adaptation does not always lead to learning.

If you're evaluating #exoskeletons (or reading studies that do), this may be the single most important thing to do 👇🏽 Before jumping to conclusions or overgeneralizing, ask yourself: 🤔 How much time were users given to acclimate to the exo (i.e., the new tool)? A short 🧵...

We’re looking forward to a great lineup of talks, panels, and posters tomorrow at #ICRA2022! Don’t miss our workshop on Online Machine Learning-Based Control of Lower Limb Exoskeletons. More info: epic.gatech.edu/icra2022/

@StevenHCollins @Nature The bar has been raised for those ME104 projects!

Applications are now open for the X Scholarship Program! This scholarship will provide 3 years of funding for an incoming EPIC Lab PhD student to work on #WearableRobotics and #AI in collaboration with @Theteamatx (formerly known as Google X). More info: tinyurl.com/39t2sybe

Out now in TMRB! We used a #deeplearning framework for estimating sagittal hip #moments during level ground/ramp/stair walking using wearable sensors. Check it out here: tinyurl.com/5e9eta74 From @dean_molinaro, @inseungkang, @jcamargo_co, @MatthewGombolay, and Aaron Young

Our recent study, "Coupled exoskeleton assistance simplifies control and maintains metabolic benefits: a simulation study", has now been published via PLOS ONE! Full text here: pubmed.ncbi.nlm.nih.gov/34986191/

A physicist, biomedical engineer, soft-goods prototyper and skateboarder walk into a lab — and they're all roboticists. On #WomensEqualityDay, X's @KatZealand shared why we need more diverse perspectives in robotics to build a future that includes everyone linkedin.com/pulse/robotics…

@shigekzishihara cambridge.org/core/journals/… thank you for letting me know the link was incorrect, here is a public link

@p_franks_ The link in your 1st tweet of this thread (link to the paper) seems accessible inside the Stanford. Is there public link to your gr8 paper(s)? Thanks.

Changes in kinematics and muscle activity indicate that single-joint assistance indirectly assisted muscles at other joints, such that the improvement from whole-leg assistance was smaller than the sum of its single-joint parts.

Looks like I accidentally listed a broken link! Here’s the article: cambridge.org/core/journals/…

#OpenAccess paper in Wearable Technologies @CUP_SciEng! "Comparing optimized exoskeleton assistance of the hip, knee, and ankle in single and multi-joint configurations." …w-cambridge-org.stanford.idm.oclc.org/core/journals/… Whole-leg assistance reduces metabolic cost of walking by 50% relative to unassisted

Two-joint assistance reduced the cost of walking by between 33 and 42%, with the largest from hip-ankle assistance (N = 3). Optimized torque profiles differed between single- and multi-joint assistance. Good starting points for designing controllers for mobile exoskeletons.

Hip-only and ankle-only assistance reduced the metabolic cost of walking by 26 and 30% relative to walking in the device unassisted, confirming both joints are good targets for assistance (N = 3). Knee-only assistance reduced cost of walking by 13% (N = 3).

I'm looking to bring on PhD and Master's students for Fall 2022 (or earlier). If you are interested in developing assistive robotics for people with mobility impairments, please reach out! Examples of past projects on my site: shepherdlab.sites.northeastern.edu

@inseungkang @dawit_lee @dongho__park @GT_EPIC_Lab @MEGeorgiaTech @GTrobotics @ScottUhlrich Multi-channel as in the sum of multiple EMG signals, each from a different muscle? That sounds like a good strategy, maybe with weights from PCSA. Problem could be underestimating passive muscle forces from lengthening. Might need an offline OpenSim model to inform.

@p_franks_ @dawit_lee @dongho__park @GT_EPIC_Lab @MEGeorgiaTech @GTrobotics @ScottUhlrich If you were to use an EMG as a proxy for joint loading (not joint moment), probably the best method would be to integrate a multi-channel approach?

We are trying to optimize our knee exoskeleton assistance profile for downhill walking. Any thoughts on what cost function we should explore? @dawit_lee @dongho__park @GT_EPIC_Lab @MEGeorgiaTech @GTrobotics

PhD is complete!