Contact
Hello and welcome!

My name is Scott and I aspire to build intelligent robots. I am particularly interested in developing robots that are resourceful in that they learn to exploit their innate dynamics to increase performance. I am endlessly motivated by the effectiveness and beauty of animal control, and the often elusive computational principles that underly it.

Presently, I am a PhD candidate in the Computer Science Department at UMass Amherst working under the supervision of Rod Grupen and Andy Barto. I have the privilege of working in two superb research labs: the Laboratory for Perceptual Robotics (LPR) and the Autonomous Learning Laboratory (ALL). I am generously supported by a NASA GSRP Fellowship from Johnson Space Center.

I mostly work with the uBot-5 and uBot-6, both dynamically stable mobile manipulators developed in the LPR. Currently, I am developing sample-efficient algorithms for learning dynamic motions that help stabilize the system after significant impacts, or to get back up after falling. This work involves an extension of Bayesian policy optimization to problems where the variance of performance is not constant across the policy space (a property present in many interesting control problems). Please email me if you would like to learn more and I will gladly share the details.

Over the last two summers I have also worked on a few projects at Johnson Space Center. These projects include developing controllers for getting up from the ground in moon gravity, drill manipulation using Robonaut 2 (R2), and development of prototype zero gravity body orientation strategies for R2 in simulation.

If you would like to learn more, please drop me an email or check out my publications, videos, or CV.