Our mission is to develop wearable robots that improve human mobility. At present, we are studying ways to improve stability and energy efficiency for individuals whose strength and coordination have been affected by amputation, stroke, or aging using robotic prostheses and exoskeletons. We believe that appropriate mechanical assistance can not only restore function, but can enhance performance beyond typical human limits.

Research Approach

Coupling theory, design and experiment. We use a combination of methods, each with different strengths and weaknesses, to obtain a complete understanding of robot-assisted locomotion. Experiments on humans allow direct measurement of human performance, discovery of new phenomena, and assessment of cooperative robot effectiveness. Simple mathematical models are intellectually accessible, allow high-throughput computational techniques such as optimization, and help inspire experimental studies. Mechanical design translates abstract ideas into practical devices for probitive experiments. We feed results from each research stream into its complements, lending depth to empirical insights, improving relevance of simulation models, and generating better assistive robots.

Experimental infrastructure development. We are developing versatile laboratory robots to facilitate scientific study of robot-assisted locomotion and to accelerate co-robot development. Rather than spend years designing and refining autonomous devices that only test a single proposed function, we are developing tethered laboratory tools that sacrifice autonomy for exceptional versatility and performance. These tools enable systematic studies of a wide range of mechanical and control functions in a single platform, allowing rapid, early-stage evaluation of proposed interventions without having to build specialized devices for each test.

Advancing core technologies. We are also developing several basic technologies to improve the effectiveness of wearable robots, with active research on control, rehabilitation techniques, actuation, and human-robot interface design. To improve energy efficiency and stability for human users, we are designing improved co-robot control strategies using control theory, model optimization, and experimental techniques. For improved range and battery size in such autonomous devices, without sacrificing controllability, we are developing novel, energy-efficient actuators. In order to apply appropriate mechanical assistance without interfering with natural human motions, we are developing lightweight and minimally-restrictive exoskeleton interfaces and multi-actuated prostheses.


Our laboratories are located in B2 Scaife Hall, a 450 ft2 space for mechanical design, bench-top testing, and pilot experiments; B6B Scaife Hall, a 300 ft2 space for treadmill gait collections; 1324 Wean Hall, an 800 ft2 space for biomechanics data collections shared with Hartmut Geyer, and B16 Scaife Hall, a 900 ft2 office space for graduate students shared with Carmel Majidi and Koushil Sreenath. We also conduct biomechanics data collections in the CMU Motion Capture Labs, with collaborators Chris Atkeson and Jessica Hodgins. Our students discuss research ideas at the weekly CMU Bipedal Locomotion Seminar, with participants from Carnegie Mellon, the University of Pittsburgh, and Disney Research Pittsburgh. We are active in the Dynamic Walking community, helping to orgainize and host annual conferences. We currently collaborate with Peter Adamczyk (University of Wisconsin, Madison), Scott Delp (Stanford University), David Morgenroth (Seattle Veterans Administration Hospital), Greg Sawicki (North Carolina State University), Kat Steele (University of Washington), Manoj Srinivasan (Ohio State University), and various companies.

Research Overview Video

February 2016 Seminar at Stanford