Abstract

Detecting the presence of a target within a field, is a long-standing problem of interest in a variety of applications from environmental to military. A framework is considered in which a single autonomous underwater vehicle (AUV), physically samples a decaying separable field generated by the target, to determine the target location. The overarching metric of interest is the three fold trade-off between the sampling and computational complexity of target detection, the cost of navigation for the AUV subject to a low detection error probability. The non-triviality of this problem stems from the emphasis on detection error probability rather than field reconstruction error and the inclusion of navigational cost as a performance metric of interest. In the absence of these two aspects, standard low-rank matrix completion theory is order-optimal in terms of sample complexity for field reconstruction. An active (adaptive) sampling and …

Date

2014

Authors

Sunav Choudhary, Dhruva Kartik, Naveen Kumar, Srikanth Narayanan, Urbashi Mitra

Conference

2014 52nd Annual Allerton Conference on Communication, Control, and Computing (Allerton)

Pages

109-115

Publisher

IEEE