high performance computing on graphics processing units: hgpu.org

This paper provides a wide perspective of the potential applicability of Graphical Processing Units (GPUs) computing power in robotics, specifically in the well known problem of 2D robotic mapping. There are three possible ways of exploiting these massively parallel devices: I) parallelizing existing algorithms, II) integrating already existing parallelized general purpose software, and III) making use of its high computational capabilities in the inception of new algorithms. This paper presents examples for all of them: parallelizing a popular implementation of the grid mapping algorithm, using a GPU open source linear sparse system solver to address the problem of linear least squares graph minimization and developing a novel method that can be efficiently parallelized and executed in a GPU for handling overlapping grid maps in a mapping with local maps algorithm. Large speedups are shown in experiments, highlighting the importance that this technology could have in robotic software development in the near future, as it is already doing in many other areas.