BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20111116T011500Z
DTEND:20111116T030000Z
LOCATION:WSCC North Galleria 2nd/3rd Floors
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: This poster describes a method for parallelizing sampling based motion planning algorithms. Our method uses subdivision of the configuration space to achieve scalability.  We subdivide the space such that graphs or trees are constructed in each subdivision using any sequential sampling-based motion planning algorithm. Graphs or trees in each subdivision are later connected to form a roadmap. =0ABy subdividing the space, we greatly reduce the overhead of inter-processor communication, a critical drawback to scalability in existing parallel motion planning work.=0A=0AAn important consequence of our approach is that it guarantees scalability. It is also general enough to handle variety of planning schemes. We compare our approach to two other existing parallel algorithms and demonstrate that our approach achieves better and scalable performance. Our preliminary results show almost linear scalability to hundreds of processors on two massively parallel =0Amachines one of which is a petascale machine at Lawrence Livermore National Lab.
SUMMARY:From Days to Seconds: Scalable Parallel Algorithm for Motion Planning
PRIORITY:3
END:VEVENT
END:VCALENDAR