History of the Robotic Ankle-Foot Prosthesis Emulator Project

I first started thinking seriously about versatile testbeds for experiments on wearable device function following conversations with Peter Adamczyk in the fall of 2008. We were frustrated by the lack of flexibility we had encountered in our experiments with the Energy Recycling foot, and got to discussing Herman van der Kooij's recent work with exoskeletons driven by off-board motors through Bowden-cables. In late November of that year, we started working on designs for a similarly-actuated prosthesis, code named "Superfoot", concluding over email that it was feasible to produce ankle torques "exceeding 200 Nm" and that one could thereby conduct experiments in which the device "perform[s] negative, zero, and three levels of positive work during push-off". Unfortunately, neither of us had sufficient time or funds to build a prototype, and so our CAD files laid dormant until the fall of 2009 when I began to apply for jobs. After discussing it with Peter, I proposed in my research statement that an ankle-foot prosthesis testbed would be my first project should I secure a faculty position.

Upon joining Carnegie Mellon in September of 2010, Josh Caputo and I immediately began hardware development. Within a few months, we had hardware working well enough for walking trials. Over the next few years many people helped to improve the platform, in particular Michiel Plooij, Soongeun Cha, and Myunghee Kim. We soon began expanding the scope of the platform by designing new end-effectors, including an ankle-foot orthosis, a two degree of freedom ankle prosthesis, and exoskeletal and prosthetic knees, with an eye towards a complete lower-limb system. Having an open and collaborative philosophy about research, we began discussing our results with colleagues immediately, including presentations at Dynamic Walking in the summer of 2011, presentations at the annual meetings of the Canadian Society of Biomechanics and American Society of Biomechanics in the summer of 2012, and several presentations in the summer of 2013. We finally published detailed descriptions of the approach and platform, along with sufficient CAD designs, component lists, and Matlab code for others to reproduce our efforts, in the Journal of Biomechanical Engineering in the fall of 2013.

As it turns out, several other researchers have pursued the idea of robotic systems for simulating wearable devices over the years. In writing our journal article, we discovered that this concept had been put forth by Woodie Flowers and Robert Mann at MIT in 1977, which is the first instance to our knowledge. These researchers used the terms "simulator" and "emulator" interchangeably, and others have used the term "testbed" in the same context. We now prefer the term "emulator" as it distinguishes the approach from purely computational work (simulator) and from tools for benchtop testing alone (testbed). Many laboratory tools for probing human biomechanics and neuromechanics share a similar design philosophy, such as those developed by Thomas Sinkjaer, Neville Hogan, Daniel Ferris, or Herman van der Kooij. More recently, many groups appear to be embracing the tethered emulator approach. We are very glad to see that the idea is catching on, and expect that these systems will lead to better experimental methodology, more generalizable scientific findings, and faster commercial development of wearable robots.