Human reach-to-grasp compensation with object pose uncertainty

Abstract

This paper examined how humans alter reach-to-grasp behavior to compensate for environmentally-induced object orientation uncertainty. We used a novel motion tracking framework to capture hand-object interactions, as well as a custom cylindrical object to detect contacts. Subjects were instructed to reach, grasp, and lift the object with or without vision. The orientation of the object was randomly changed on each trial. We hypothesized subjects would use a reach-to-grasp strategy that minimizes post-contact adjustments. However, our results indicate that (1) subjects are more likely to use the hand as a sensing apparatus prior to contact, and (2) the reach-to-grasp kinematics may be optimized for efficient sensing of object orientation. Our findings could provide potential solution to efficient tactile sensing for robotic hand in unstructured environment.