Feature Based Localization and Navigation for Miniature Underwater Vehicles

Period of Performance: 01/01/2015 - 12/31/2015


Phase 2 SBIR

Recipient Firm

Greensea Systems, Inc.
10 East Main Street P.O. Box 959
Richmond, VT 05477
Principal Investigator


This proposed effort will continue the development, testing, and integration of a general-purpose feature-based navigation system for miniature UUVs that provides a significant reduction in size, weight, and power (SWaP) requirements over current navigation systems and methodologies while providing the accuracy and robustness needed for viable underwater operations. This navigation system utilizes a micro mechanically scanned imaging sonar (uMSIS) as a perception sensor and a MEMS-based Inertial Navgiation System (INS). By detecting, extracting, and matching features from the sonar data, the proposed system generates a vehicle pose estimate that is directly related to the environment and uses this pose estimate to correct the inertial drift of the navigation system. As well, the system builds an online real-time map of the detected features to assist in navigating back through that environment. Our approach uses a generic, flexible, and scalable software architecture to manage the sonar sensor, build a real-time geo-corrected image, extract and match features from the image, estimate a feature-based vehicle pose, and implement a deconstructed Inertial-Simultaneous Localization and Mapping (Inertial-SLAM) algorithm optimized for light-weight embedded systems.