 Hundreds of millions of people are unable to walk just a few hundred meters because of disabilities like amputation or stroke or simply the effects of aging. Many more labor under strenuous walking conditions, for example in disaster relief. My colleagues and I want to design wearable devices that enhance these people's mobility and not just restore it but enhance it. Prosthesis that gives you superhuman balance. An exoskeleton that allows you to walk faster than before your stroke or carry a heavy pack all day without tiring. These devices should be efficient and cheap so everyone can have one. Such technology would improve mobility for all of us. Delivering on this promise is much harder than it seems. For the thousands of prototypes designed over the past century, our best handful have provided only modest improvements in performance. The problem has not been a lack of attention or investment. Rather, in the design approach itself. See, people tend to think that making a good exoskeleton requires only clever mechanical design. The real challenge is in how these devices interact with the user. Humans are highly complex. No two of us are the same and we're constantly changing. So designed by intuition or tinkering is ineffective. One way to overcome this problem is to bring humans more fully into the design process. By continually testing actual users to find features that help them. Of course that could be laborious. So we need tools to speed and systematize the process of discovering the best features. One such tool that we've developed we call a universal device emulator. Users wear a lightweight prosthesis or exoskeleton tethered to powerful off-board motors and computers. These systems are not mobile, but they are exceptionally versatile, enabling tests of designs that can later be made and taken home. Emulators allow the user to experience a wide range of devices without the need to build new specialized hardware. It's like virtual reality, but for your legs. Any design, real or hypothetical, can be programmed into the emulator, allowing measurements of human response in days rather than years. Taking this approach a step further, we can even automate the process of iteratively guessing at new designs, tightening the loop to minutes. We call this human in the loop optimization, in which device features are systematically varied during use so as to maximize human performance. The user drives design directly. We recently demonstrated this approach using an algorithm that adapts alongside the user during walking to determine individualized patterns of assistance for an exoskeleton worn on one ankle. Optimized assistance improved human energy economy by 24% more than any other device to date. With this approach, it's relatively easy to make emulators for different objectives, like a multi-actuated prosthetic foot for improving balance on uneven terrain, or an exoskeleton to offload injured joints during running. We can then try to optimize many aspects of human performance. Commercial emulators, like this one from a spin-off of my lab, could soon be found in storefronts and clinics where they will identify specifications for custom devices that are then fabricated and used in daily life. Those custom devices should be as efficient and cheap as possible, so we finally do need clever mechanical design. For example, this ankle exoskeleton we developed uses a mechanical clutch in spring with no heavy motors or batteries. It uses no energy itself. It reduces the energy cost of human walking. Unpowered assistance techniques like this, discovered by emulators, lend themselves to elegant and inexpensive products. To extend what's possible in such products, we'll need better robotics hardware too. This electrically controllable clutch we've developed weighs less than a penny and could run off of the battery in your smartphone for decades, enabling some very exciting new system designs. Now, everything I've been telling you about today is right on the cutting edge. Until just three years ago, no device had succeeded in making walking easier than normal. Now, with techniques like emulation and human and loop optimization, we're finally starting to figure out what your device should do to assist you. With efficient and low-cost hardware, we can finally start to make products that are widely accessible. The future of human mobility is bright. If we recognize the real challenges and invest wisely, we have the opportunity to create affordable, customized, assistive devices changing life for millions of people and allowing all of us to overcome the limits of our current bodies. Thank you.