A framework of concurrent navigation and seabottom targets detection using acoustic sensors on AUVs

Abstract

The use of Autonomous Underwater Vehicles (AUVs) for Mine Counter Measures (MCMs) is an area of active recent research. The excellent mobility of AUVs allows for multi-aspect sonar view of the targets for improved detection, tracking, and classification. However, the uncertainty of the platforms and associated target localization degrade the detection ability by AUVs. Furthermore, the weak signals from buried targets is another severe problem making detection by a single measurement impossible. A new acoustic sensing framework is proposed for detection and tracking seabottom targets based on the track-before-detect (TBD) technique. In contrast to the traditional methods, TBD tracks possible targets before the detection is declared. There are several advantages by using this framework: (1) Compared to the traditional method of detection followed by tracking, higher detection probability is achieved for dim target detection due to the integrated detection metric of TBD. (2) Compared to the multiple hypothesis tracker (MHT) method, instead of searching a diverge hypotheses tree, TBD speeds up the searching and decreases the computational load, which makes onboard implementation feasible. (3) The stochastic models of uncertainties of targets and AUVs are based on the Bayesian framework, and thus, it is easy to apply various recursive estimators such as the Kalman filter or particle filter for tracking individual targets as well as the AUV platforms. Results of a successful application of this method in the GOATS2002 experiment are demonstrated. [Work supported by ONR and NATO Undersea Research Centre.]