• DocumentCode
    3019152
  • Title

    Akers´s wavefront planner — One of the fastest stencil-based path planners on FPGAs

  • Author

    Schmidt, Martin ; Fey, D.

  • Author_Institution
    Dept. Comput. Sci., Friedrich-Alexander-Univ. Erlangen-Nuremberg, Nuremberg, Germany
  • fYear
    2012
  • fDate
    5-7 Dec. 2012
  • Firstpage
    1
  • Lastpage
    6
  • Abstract
    The efficient and fast processing of path planning algorithms based on stencil codes is a great challenge. On the one hand, these iterative approaches are advantageous because of their regular processing pattern and the characteristic to avoid problems of other path planning methods, like local minima in potential field algorithms. On the other hand, they are computational- and data-intensive because of their high number of required iterations. By comparing several algorithms based on stencil codes, we found one of the fastest and most efficient path planners for realization on FPGAs, Akers´s wavefront planner. A processing of 33 maps per second with a resolution of 1024×1024 is possible on a midsize Virtex-5 FPGA achieved through a column-based processing scheme combined with an efficient internal data storage.
  • Keywords
    collision avoidance; field programmable gate arrays; iterative methods; matrix algebra; mobile robots; path planning; Akers wavefront planner; Virtex-5 FPGA; collision-free path; column-based processing scheme; computational-intensive; data-intensive; internal data storage; iterative approaches; local minima; path planning algorithms; regular matrices; robot systems; stencil codes; stencil-based path planners; Clocks; Computer architecture; Field programmable gate arrays; Microprocessors; Path planning; Registers; Robots; FPGA; path planning; stencil codes; wavefront;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Reconfigurable Computing and FPGAs (ReConFig), 2012 International Conference on
  • Conference_Location
    Cancun
  • Print_ISBN
    978-1-4673-2919-4
  • Type

    conf

  • DOI
    10.1109/ReConFig.2012.6416727
  • Filename
    6416727