• DocumentCode
    1865201
  • Title

    The laser line object detection method in an anti-collision system for powered wheelchairs

  • Author

    Huang, Hai ; Fernie, Geoff R.

  • Author_Institution
    Res. Group, Toronto Rehabilitation Inst., Ont., Canada
  • fYear
    2005
  • fDate
    28 June-1 July 2005
  • Firstpage
    173
  • Lastpage
    177
  • Abstract
    The residents in long term care facilities with cognitive impairment and mobility disability need an anti-collision system on their powered wheelchairs to prevent them from causing other seniors to fall. Because of the severe consequence of falling, the detection method of the anti-collision system must be very reliable. However, many object detection techniques tend to miss targets that are unfavourably oriented or have certain surface properties. This research evaluated an uncommon method: laser line object detection (LLOD). The LLOD system projects an invisible infrared laser line onto the ground, and reads the resulting image via a camera. By analyzing the laser line in the image, the system can identify whether objects are in the target area. A pilot LLOD system was designed and installed on a powered wheelchair. The results of the evaluation experiments showed that the LLOD system can detect almost all obstacles with different orientations and materials, and produced a high detection rate on favourable flooring surfaces.
  • Keywords
    collision avoidance; computer vision; geriatrics; handicapped aids; object detection; anti-collision system; cognitive impairment; laser line object detection; laser line object detection method; mobility disability; powered wheelchairs; Image analysis; Infrared detectors; Infrared sensors; Intelligent sensors; Laser noise; Object detection; Power lasers; Power system reliability; Sonar detection; Wheelchairs;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Rehabilitation Robotics, 2005. ICORR 2005. 9th International Conference on
  • Print_ISBN
    0-7803-9003-2
  • Type

    conf

  • DOI
    10.1109/ICORR.2005.1501078
  • Filename
    1501078