CAREER: Toward Robust Multi-Vehicle Multi-Scalar Underwater Robotic Navigation - A Career Development Plan

点击翻译按钮获取中文摘要

The objective of this career development plan is to investigate newprobabilistic methods for simultaneous localization and mapping (SLAM)that will improve the precision and scale of robotic mapping in oceanscience. The proposed methodology will scale to multipleheterogeneous vehicles, allow for extended exploration over long timedurations and over multiple spatial scales, and be robust to thechallenging limitations of the underwater environment. Many land/airSLAM methods are largely inapplicable underwater because of (a) a lackof point features of unstructured seafloor, and (b) the rapidattenuation of electromagnetic, optical, and acoustic radiation incomparison to land, air, and space. The proposed multi-vehiclemulti-scalar SLAM navigation framework overcomes these challenges byfusing information from three disparate technologies: 1) real-timevision-based seafloor navigation, 2) inertial navigation systems, and3) acoustic modem-based communication, to create a flexible navigationframework that allows for inter-nodal ranging and data sharing amongheterogeneous nodes. By leveraging the perception and localizationcapability of neighboring vehicles via a (low-bandwidth) distributedestimation framework, the proposed methodology will provide acooperative navigation framework that will yield improved precisionand scale in underwater robotic mapping for ocean science.