A much more detailed look at precision manipulation workspace is now available through our paper in IEEE Transactions on Biomedical Engineering (TBME). Below is an example figure from the paper. This figure shows typical finger positions relative to the object in both two- and three-fingered manipulation of a small object in the fingertips.
The thumb position appears to stay fairly constant, while the position of the other fingers is significantly different depending on whether two or three fingers are used. This data could be used, for example, to anticipate natural finger placement when designing devices which are held in the fingertips during use. Please see the paper for more details.
My paper for IROS 2014 discusses human fingertip usage during precision manipulation. Specifically, participants manipulated a small spherical object with 2 or 3 fingertips (thumb and index or thumb, index, and middle fingers). Participants explored their workspace, and the resulting object and fingertip locations were recorded using a magnetic tracker. The results show quite substantial use of the middle finger on its side surface in the 3-finger case, as well as a shift in usage of the index finger. The index finger is used more on its lateral surface in the two finger case than in the three finger case.
Overall, the results suggest that finger side surfaces could be useful for robotic manipulators designed for precision manipulation. It is interesting that the middle finger is used in a direction that it can’t directly actively apply force, rather the passive stiffness of the finger is often used to support the object during manipulation.
The paper can be seen here.
I recently attended the ICRA 2013 conference in Karlsruhe, Germany. A few highlights are below:
My paper, Finding Small, Versatile Sets of Human Grasps to Span Common Objects, was a finalist for Best Manipulation Paper. I didn’t win, but they still gave me a nice certificate:
I am honored by this recognition for my work with Thomas Feix and Aaron Dollar.
I also got to help out with the OpenHand workshop, led by Raymond Ma and Lael Odhner. It was fun to teach a bunch of people about how our lab’s underactuated hands work, and how they can be fabricated.
A journal paper analyzing the grasping of two housekeeper and two machinist subjects over about 30 hours was accepted to Transactions on Haptics, and a pre-print version is now available. Two sample figures are shown below, but please see the full paper for much more detailed analysis.
Snapshots of the most common grasps for the four subjects.
My paper, Finding Small, Versatile Sets of Human Grasps to Span Common Objects, was recently accepted for publication in the proceedings of ICRA 2013. It was written with Aaron Dollar and Thomas Feix.
A novel grasp span metric is used to assess the versatility of a small set of grasps in handling a wide range of objects. We used over 9000 grasps over 19 hours of video from four subjects to generate small sets of versatile grasps using this metric. The results for all subjects are below:
A journal paper submitted to Transactions on Haptics was recently accepted (Thank you to my co-authors Raymond Ma and Aaron Dollar). It is titled A Hand-Centric Classification of Human and Robot Dexterous Manipulation, and discusses a general framework for classifying manipulation of both human robotic systems.
For the moment, please see the preliminary ICORR 2011 paper, but the full journal paper should be available soon. A few quick snapshots are provided to get a general idea of the journal paper content:
A few important terms are used to classify manipulation tasks.