Error modeling and calibration of exteroceptive sensors for accurate mapping applications

Robust, scalable simultaneous localization and mapping (SLAM) algorithms support the successful deployment of robots in real-world applications. In many cases these platforms deliver vast amounts of sensor data from large-scale, unstructured environments. These data may be difficult to interpret by end users without further processing and suitable visualization tools. We present a robust, automated system for large-scale three-dimensional (3D) reconstruction and visualization that takes stereo imagery from an autonomous underwater vehicle (AUV) and SLAM-based vehicle poses to deliver detailed 3D models of the seafloor in the form of textured polygonal meshes. Our system must cope with thousands of images, lighting conditions that create visual seams when texturing, and possible inconsistencies between stereo meshes arising from errors in calibration, triangulation, and navigation. Our approach breaks down the problem into manageable stages by first estimating local structure and then combining these estimates to recover a composite georeferenced structure using SLAM-based vehicle pose estimates. A texture-mapped surface at multiple scales is then generated that is interactively presented to the user through a visualization engine. We adapt established solutions when possible, with an emphasis on quickly delivering approximate yet visually consistent reconstructions on standard computing hardware. This allows scientists on a research cruise to use our system to design follow-up deployments of the AUV and complementary instruments. To date, this system has been tested on several research cruises in Australian waters and has been used to reliably generate and visualize reconstructions for more than 60 dives covering diverse habitats and representing hundreds of linear kilometers of survey. © 2009 Wiley Periodicals, Inc.