• DocumentCode
    1635959
  • Title

    Human pointing as a robot directive

  • Author

    Abidi, Samina ; Williams, McKay ; Johnston, Benjamin

  • Author_Institution
    Social Robot. Studio, Univ. of Technol., Sydney, NSW, Australia
  • fYear
    2013
  • Firstpage
    67
  • Lastpage
    68
  • Abstract
    People are accustomed to directing other people´s attention using pointing gestures. People enact and interpret pointing commands often and effortlessly. If robots understand human intentions (e.g. as encoded in pointing-gestures), they can reach higher levels of engagement with people. This paper explores methods that robots can use to allow people to direct them to move to a specific location, using an inexpensive Kinect sensor. The joint positions of the pointing human´s right arm and hand in 3D-space are extracted and used by the robot to identify the direction of user´s pointing gesture. We evaluated the proposed approach on a PR2 robot whose task was to move to a location that a human pointed to on the ground. This method enables the robot to follow human pointing gestures on the fly and in real-time. It will be deployed on a PR2 in the wild in a new building environment where the robot will be expected to interact with people and interpret their human pointing behaviors.
  • Keywords
    gesture recognition; human-robot interaction; mobile robots; position control; 3D-space; Kinect sensor; PR2 robot; building environment; direction identification; hand position; human intention understanding; human pointing behavior; human pointing gesture; pointing command; pointing human right arm position; position extraction; robot directive; robot movement; service robot; user pointing gesture; Cameras; Robot vision systems; Skeleton; Thumb; Vectors; Human Robot Interaction; directing attention with gestures; human pointing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Human-Robot Interaction (HRI), 2013 8th ACM/IEEE International Conference on
  • Conference_Location
    Tokyo
  • ISSN
    2167-2121
  • Print_ISBN
    978-1-4673-3099-2
  • Electronic_ISBN
    2167-2121
  • Type

    conf

  • DOI
    10.1109/HRI.2013.6483504
  • Filename
    6483504