SBIR-STTR Award

This development would improve off-GPS navigation performance for unmanned and manned underwater vehicles, including UUVs, ,AUVs, USVs and more and with automatic, real-time Simultaneous Localization and Mapping (SLAM)navigation aiding using matching of terrain features to prove corrections that reduce the drift in the vehicle's INS thereby improving mission performance while reducing the need for costly post-acquisition processing, or re-survey. The development will be based on Oceanic Imaging Consultants' (OIC's) commercially available product, well established in the industry, that performs manual feature matching for sidescan and swath bathymetry data utilizing a SLAM algorithm in post-acquisition mode. The tasks proposed in this effort are 1) automation of feature detection, extraction and matching, 2) optimization of matched feature selection to minimize outlier-induced noise, and 3) modification or replacement of the current working SLAM algorithm to run incrementally, and in real-time. This would solve a non-trivial problem for existing government programs, and is supported enthusiastically by existing and future OIC customers. OIC's proposes that the phase I desktop and testing effort be followed by a 6 month effort of in-depth testing and refinement, with collection of purpose-built test data. The result would be algorithms ready for implementation and documentation in production code, interfaced to the Navy's Unmanned Maritime Autonomy Architecture (UMAA), in preparation for trials on Government UUVs and commercial vehicles.

Benefit:
Development of a robust, real-time Simultaneous Localization and mapping (SLAM) solution for aiding the navigation of UUV's AUV's and USV's operating in GPS-denied environments would be game-changing for the seafloor mapping and Navy mission communities. Without GPS updates, current INS navigation solutions available to these vehicles all suffer drift over time, resulting in positional uncertainties which can compromise data usefulness and mission completion. Addition of a SLAM-based feature matching solution to minimize drift and aid correct vehicle navigation would be of enormous benefit. As both NAVY and commercial UUV/AUV/UxV fleets are growing daily, the market is expanding continuously. Where single post-processing centers might in the past have serviced multiple vehicles, moving the algorithm from the office to the field easily increase the market size by an order of magnitude. Existing customers, both commercial and governmental, have expressed their interest in supporting and acquiring this technology, if successful. With sensors on Remotely Operated Vehicle (ROVs) and Unmanned Surface Vehicles (USV) now exceeding in resolution their current possible positioning accuracy, an additional market of even greater size has presented itself. Even commercial, military and academic deep-tow operations could benefit, as no commercially available solution exists to position a towed vehicle at accuracy greater than that of current imaging systems.

Keywords:
UUV navigation, UUV navigation, Feature-based navigation, real-time SLAM

This effort seeks to improve off-GPS navigation performance for underwater vehicles, both manned and unmanned, including UUVs, AUVs, ROVs and manned subs, with automatic, real-time feature-based navigation known as SLAM - Simultaneous Localization and Mapping. SLAM navigation aiding uses matching of terrain features to prove corrections that reduce the drift in the vehicles Inertial Navigation System (INS), thereby improving mission performance while reducing the need for costly post-acquisition processing, or re-survey. Then development will be based on Oceanic Imagining Consultants' (OIC's) commercially available product, well established in the industry, that performs manual feature matching for sidescan and swath bathymetry data using a SLAM algorithm tin post-acquisition mode. The tasks outlined in this proposal are 1) getting the current or updated SLAM algorithm to run in real-time, 2) addition of support for processing of Forward-Look Sonar data in CleanSweep, 3) automation of feature-detection, extraction and matching, and 4) optimization of matched feature selection to minimize outlive-induced noise in re-estimation of navigation trajectory. This would solve a non-trivial problem for existing government programs, and is supported enthusiastically by existing and future OIC customers on the commercial side. The result will be algorithms ready for implementation and utilization in production code, interfaced to the Navy's Unmanned Maritime Autonomy Architecture (UMAA), in preparation for trials on Government UUV's and commercial vehicles.

Benefit:
Development of a robust, real-time feature-based navigation solution using stable, Simultaneous Localization and Mapping (SLAM) techniques for aiding in the in-mission navigation of UUV's, AUV's, ROV's and manned underwater vehicles operating in GPS-denied environments would be game-changing for the seafloor mapping and Navy mission communities. Without GPS updates, current INS navigation solutions available to these vehicles all suffer drift over time, resulting in positional uncertainties which can compromise data usefulness and mission completion. Addition of a SLAM-based feature matching solution to minimized drift and aid correct vehicle navigation would be of enormous benefit. as both USN and commercial UUV/ROV/AUV/UxV fleets are expanding the market is growing.

Keywords:
Factor Graphs, SLAM, multi-modal matching, FEATURE DETECTION, Feature-based navigation