A detailed study was conducted to prove the concept of an iterative approach to single transponder navigation for REMUS Autonomous Underwater Vehicles (AUVs). Although the concept of navigation with one acoustic beacon is not new, the objective was to develop a computer algorithm that could eventually be integrated into the REMUS architecture. This approach uses a least squares fit routine coupled with restrictive geometry and simulated annealing vice Kalman filtering and state vectors. In addition, to provide maximum flexibility, the single transponder was located on a GPS equipped surface ship that was free to move instead of the more common single bottom mounted beacon. Using only a series of spread spectrum ranges logged with time stamp, REMUS standard vehicle data, and reasonable initial conditions, the position at a later time was derived with a figure of merit fit score. Initial investigation was conducted using a noise model developed to simulate the errors suspected with the REMUS sensor suite. Results of this effort were applied to a small at sea test in 3,300 meters with the REMUS 6000 deep water AUV. A more detailed test was executed in Buzzard's Bay, Massachusetts, in 20 meters of water with a REMUS 100 AUV focusing on navigation in a typical search box. While deep water data was too sparse to reveal conclusive results, the Buzzard's Bay work strongly supports the premise that an iterative algorithm can reliably integrate REMUS logged data and an accurate time sequence of ranges to provide position fixes through simple least squares fitting. Ten navigational legs up to1500 meters in length showed that over 90% of radial position error can be removed from an AUV's position estimate using the STRONG algorithm vice dead reckon navigation with a magnetic compass and Doppler Velocity Log alone (DVL)
展开▼
