Exoskeleton Hardware and Control Development: To enable much of my work with exoskeletons as a grad student, I built a bilateral ankle exoskeleton testbed system. In addition to the hardware, I developed the high and low-level control [1] for my own work in applying impedance control [2,3] and for projects using speed adaptive proportional EMG [4] using neuromuscular-based models (in review). As a post-doc, I lead teams, set design criteria, and develop soft exosuit hardware and controls for targeting ankle, knee, and hip in healthy and clinical populations.


Muscle-tendon neuromechanics of human-robot interaction by using new sensing tools: This body of work focuses on an understanding of muscle tendon dynamics and how that understanding can be applied to improving wearable robotic assistance. An understanding of muscle dynamics is crucial as the muscle is what uses energy and generates the force, yet directly measuring muscle dynamics is difficult. While previous work had provided insights into how exoskeletons affect whole body and individual joint biomechanics, my PhD work specifically targeted gaining a better understanding of muscle dynamics [2,5]. No prior study had directly measured how exoskeletons affected muscle fascicle dynamics during exo-assisted walking. I used ultrasound imaging to visualize and measure the kinematics of the plantarflexor muscles of subjects walking with multiple levels of exoskeleton assistance (Fig. 2). We showed that assistance can alter muscle contraction dynamics, impact the muscle’s ability to generate force, and thus ultimately affects the user’s response. These findings are important as they help explain why and how individuals respond to exoskeleton assistance. This knowledge can be applied to improve models of human-robot interaction, aide in development of new assistance strategies that target muscle-tendon (MT) dynamics and can also be expanded into a better understanding of changes occurring during intervention with clinical populations.

A knowledge of MT dynamics can also inform how we should apply exosuit or exoskeleton assistance. Using my knowledge of MT dynamics, at Harvard I developed a strategy of designing individualized assistance profiles that account for changes and variance among different subjects [5,6]. First, I developed techniques for measuring muscle dynamics in real-time and showed that we can capture changes in muscle dynamics related to the individual and environmental demand. Finally, we demonstrated that this information can be used to tailor assistance profiles and maximize user benefit.

     I am at the forefront of developing and implementing new technologies to help answer previously unanswerable questions as exemplified by my work with ultrasound imaging to evaluate exo-assisted walking, development of image processing techniques, and use of ultrasound in developing assistive strategies. Another example of where I have looked for new tools to is the recent grant that I took the lead in writing: NSF DARE #2019580: “Quantifying Reductions in Musculoskeletal Loading due to Soft Exosuit Assistance using Shear Wave Tensiometry” (Conor Walsh, PI) in collaboration with Dr. Darryl Thelen (Wisconsin). The incorporation of muscle-tendon neuromechanics will continue to be an important research area for understanding healthy and clinical biomechanics and for the tailoring of assistance.


Multi-Scale Mechanics

Figure: Multi-scale analysis to understand effects of wearable robotics on human neuromechanics. 


Muscle Gif

Figure: B-mode US image sequence of calf muscles (Soleus and Gastroc) with automated detection of heel strike and muscle shortening


Adaptive wearable robotics for real-world conditions: I am interested in developing wearable devices that adapt to real-world environmental and task demands such as variable walking speed and/or inclined surfaces. As part of my PhD, we investigated how walking speed impacts the effectiveness of passive elastic ankle exoskeletons and showed that passive devices were effective at providing metabolic benefit at multiple speeds [3]. Interestingly, the exo stiffness that provided the best metabolic benefit changed only slightly from slow to fast walking speed and suggest that simple, yet effective passive designs might be possible. Furthermore, by investigating several walking speeds and assistance levels, we were able to show that a general strategy of minimizing muscle activation of the assisted muscle, while limiting co-contraction of opposing muscles leads to the best whole-body energetic improvement.

     During my PhD, I also led a biomechanics study to inform exoskeleton design requirements for incline walking and running [7]. We showed, for example, how during incline walking and running, joint power shifts between the ankle and hip. These analyses were organized into a framework for thinking about how wearable devices could interact with the person in these changing environmental demands (Fig. 3). Recently at Harvard, I applied this knowledge and showed that we can capture the changes in muscle dynamics related to incline walking using ultrasound imaging and demonstrated a proof-of-concept for how this information can be used to tailor assistance profiles and maximize user benefit [6]. Much like designing for the individual, the design for dynamic environments will continue to be an exciting area of research.




Incline Walking




Improving clinical gait with wearable robotics: My work has also focused on understanding how we might use technologies for providing enhanced rehabilitation and or improved functional movement for individuals with gait deficits. During my PhD, I developed a hardware and control system that adapted assistance profiles to the walking speed of stroke survivors for improving post-stroke gait [4]. This area of research also incorporates my understanding of muscle tendon dynamics. I performed preliminary studies to measure age related shifts in muscle dynamics and show how show that we could offset these changes and improve energy use [8].

     More recently during my postdoc, I have been leading a team focused on developing the lab’s next generation of wearable robotics devices for assisting and rehabilitating clinical gait. This team is highly interdisciplinary and encompasses graduate students, post docs, staff engineers, clinicians, and functional apparel designers. The lab has had success with the development and commercialization of the mobile ankle systems, but stroke gait is heterogeneous and we find that rehabilitation techniques and assistance strategies that are effective for one individual and situation are not necessarily effective in another. Based on our knowledge of fundamental and clinical biomechanics, we developed a hip flexion exosuit for assisting gait in stroke survivors [9]. Using this system, we’ve begun investigating how assistance can be individualized such that it targets the needs of stroke survivors during overground gait in the lab [10]. Finally, our team has begun investigating how these systems can be used to improve in clinic rehab therapy. Through our collaborations with Shirley Ryan Ability Labs and Spaulding Rehab Hospital, we have begun testing with our exosuits with outpatient and inpatient stroke survivors and their therapist [11]. To enable this, our team has begun developing ‘enhanced research devices’ which are designed to fill the gap between the research lab and commercial products. These systems will allow us to bridge the gap between the lab and the clinic and allow us to begin testing much earlier in the development cycle of our devices. By doing this, we can iterate more effectively and address the needs of the therapist and patient in a more effective manner and improve rehab outcomes.