We aim to develop an open-source, low-cost (i.e., hobbyist-friendly) exoskeleton. Currently we've put in exactly 30 hours into the project (during a hackathon) and have produced a single working joint as a prototype.
Our goal was to use a high RPM brushless DC motor and gear it down to useful speeds. The motor we had was about 8,000 RPM at top speed with no load. We figured that in order for the motor to be useful we would need about a 100:1 gear reduction. Our finished prototype was able to get a 50:1 gear reduction, using a cycloidal gear drive.
The reason we used a cycloidal gear was because it is not backdriveable, efficient, and offers high gear reduction in a small flat package. The downside to a cycloidal gear is that it vibrates due to the eccentric cam.
Last weekend, we spent 30 hours at OSU's HWeekend event hacking together a single joint as a prototype for what will eventually be a full-body exoskeleton.
We don't have a fully fleshed-out plan for this yet, but hopefully this will be something relatively cheap to build with tools and materials commonly available to hobbyists while still being a capable machine: fast enough to feel immaterial and strong enough to fully support a 150-pound human body.
I've always been a fan of cycloidal reductions - even with their downsides, they are an ingenious mechanism. Thanks for entering CTRL in The Hackaday Prize. Up until now, exoskeletons have been research and science fiction - I hope you keep going with this project and take it all the way to space!