Human Head Motion Modeling Using Monocular Cues for Interactive Robotic Applications

摘要

In this paper, a generalized 2D trajectory models are developed for human motion during walking and running for a user-centered Human-Computer Interaction design. The mapping of the head motion trajectories has been done through monocular cues to compute instantaneous displacement and velocity. The novelty of the work lies in using non-contact sensor only which saves from the use of complex body-contact sensors as well as improves data storage and processing. The 2D trajectory models developed are tested on 30 subjects and that show the average head motion trajectory map of a running person differs from that of the walking person on a treadmill. The quantification of head velocity values as a function of walking and running velocities is also tested which can be used to identify different locomotion velocities from head velocities.