GPU Accelerated Voxel Grid Generation for Fast MAV Exploration

Abstract-- Voxel grids are a minimal and efficient environment representation that is used for robot motion planning in numerous tasks. Many state-of-the-art planning algorithms use voxel grids composed of free, occupied and unknown voxels. In this paper we propose a new GPU accelerated algorithm for partitioning the space into a voxel grid with occupied, free and unknown voxels. The proposed approach is low latency and suitable for high speed navigation. I. INTRODUCTION Many sensors (RGB-D cameras, stereo-matching...) output dense pointclouds as measurements and need to be processed and turned into an environment model/representation for motion planning.