Matthew T. Mason
Professor, Computer Science and Robotics
Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh PA 15213
NSH A521 (Newell-Simon Hall A521)
Voice: (412) 268-8804
Fax: (412) 268-7350
I work on manipulation, mainly robotic manipulation, but in addition I have started working on manipulation by human beings. The primary locus for my work is the Manipulation Lab also known as the MLab.
Simple Hands. My primary focus for the last five years or so has been simple hands. I think simple hands are better than complex anthropomorphic hands, both for research purposes and for near-term practical use. The main reason is that the dexterity of a hand is mostly about the brain, not the hand. Our approach mixes a strong dose of machine learning into the traditional brew of AI, control and mechanical design. Here's a good look at some of our most recent stuff:
If you're curious, there are publications here. Probably the first one to read is the 2012 IJRR paper Autonomous Manipulation with a General-Purpose Simple Hand. Also see related papers from IJRR 2014 and ICRA 2014. The two papers published in ICRA 2013 and RSS 2011 each won best student paper awards. The project began with NSF funding, continued with funds from the Army Research Office and ABB, and just recently won another NSF award. This is a collaboration with MIT Professor Alberto Rodriguez and CMU Professors Mike Erdmann and Sidd Srinivasa.
Human manipulation. It is strange how little we seem to know (scientifically) about human maniplulation. It seems we know more about ape manipulation than human manipulation. A 2001 study identified 190 different ape manipulation primitives used in a single task. There hasn't been a comparable study of human behavior. Robotics research papers are full of assertions about how humans behave, often with little or no scientific basis. The immediate plan, working with CMU Professor Nancy Pollard and MIT Professor Alberto Rodriguez, is to use citizen science to analyze the vast corpus of human manipulation videos available on the web. The resulting database would support work on robotic manipulation as well as plain old scientific research on human behavior.
Older projects. For older projects, check out the Manipulation Lab music video, with a funky sound track by Eldridge Gravy and the Court Supreme:
or the Mobipulator video:
I teach 16-741 Mechanics of Manipulation annually in the Fall term. The course web page is here.
I used to teach 15-384, an undergraduate class on manipulator kinematics, control and programming. We had some fun course projects like robots throwing darts, or bowling.
I also used to teach the robotics section of Andrew's Leap, a summer program for high school students. One time we built a robot that dangled a dollar on a string from a balcony to lure humans, then dropped water balloons on them. That was 1996. Here's the video.