Using haptics to probe human contact control strategies for six degree-of-freedom tasks

• Published in: 2014 IEEE Haptics Symposium (HAPTICS) pp. 93 - 95
• Main Authors: Klingbeil, Ellen, Menon, Samir, Go, Keegan, Khatib, Oussama
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-02-2014
• Subjects: Dynamics; Educational institutions; Force; Force rendering; haptic; Haptic interfaces; Manufacturing/assembly; Perception and psychophysics; Rendering (computer graphics); Robots; Trajectory; Virtual environment modeling
• ISSN: 2324-7347; 2324-7355
• DOI: 10.1109/HAPTICS.2014.6775438

Transferring human contact-manipulation skills to robots is complicated by the combinatorial growth in contact state descriptions with object and environment complexity, with no systematic method to characterize the subset of contact states that humans actually control. In this paper, we present an approach to determine whether human subjects control specific contact states. Subjects used a six degree-of-freedom haptic interface to place a box on a plane, and insert L- and S-shaped objects into corresponding holes, while using either their (undetermined) natural strategies or explicitly controlling a pre-specified sequence of contact states. We found that the vast majority of contact states were visited for time periods of less than 40ms, which negates the possibility of human feedback control due to physiological delay limits. Next, using force, moment and velocity trajectories around contact state transitions as a metric, we found that certain states were readily discriminable across natural and explicit control strategies (~ 85%), which indicates that they were not controlled during natural motions. Less discriminable states, in contrast, were likely to have been controlled in both natural and explicit strategies. Our results suggest that humans explicitly control a small subset of contact states, and that their control strategies are reflected in local force and velocity profiles. We thus demonstrate that six degree-of-freedom haptic simulations are effective for characterizing human contact-state invariant control strategies.