OHM: GPU Based Occupancy Map Generation

Occupancy grid maps (OGMs) are fundamental to most systems for autonomous robotic navigation. However, CPU-based implementations struggle keep up with data rates from modern 3D lidar sensors, and provide little capacity extensions which maintain richer voxel representations. This article presents Homogenous Mapping (OHM), our open source, GPU-based OGM framework. We show how the algorithms can be mapped GPU resources, resolving difficulties contention obtain a successful implementation. The implementation supports many including Normal Distributions Transform-Occupancy Maps (NDT-OM), Transform-Traversability (NDT-TM), decay-rate Truncated Sign Distance Function (TSDF). A thorough performance evaluation is presented based on tracked quadruped Uncrewed Ground Vehicle (UGV) platforms UAVs, sets both outdoor subterranean environments. results demonstrate excellent improvements offline, online processing in embedded platforms. Finally, we describe OHM was key enabler UGV navigation solution entry Defense Advanced Research Projects Agency (DARPA) Subterranean Challenge, placed second at Final Event.