Synchronization for large network of marine active drifting sensors through periodic intermittent rendezvous
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
This dissertation focuses on mobile sensor network coordination for long-endurance large scale marine monitoring and surveillance that leverages ambient environmental dynamics such as those represented by Lagrangian Coherent Structure (LCS). The control approach developed steers the sensor network to establish intermittent periodic rendezvous in order to enable the network to efficiently propagate and upload information. The dissertation first offers conditions under which mobile sensors that are leveraging the environmental dynamics along adjacent periodic orbit will rendezvous in a small neighborhood intersecting with their paths. The working theoretical assumption is that the ocean currents that drive the robotics sensors can be approximated as gyres or eddy flows arranged over a grid, in which each gyre is delineated by Lagrangian coherent structures. The result is generalized from the case where sensor motion is a harmonic oscillation to that in which the robotic sensors drift periodically along the theoretical ocean circulation paths in periodic ocean circulation. ☐ Assuming that robotic sensors can interact with each other only when they are in rendezvous, this dissertation presents controllers motivated by different application challenges that regulate robotic sensors’ motion characteristics over the small time window of interaction in order to robustify their synchronous rendezvous from the first occurrence of that event and in perpetuity. The development of the control strategy for the synchronization of the robotic sensor network starts with pairwise control that aims at regulating the robotic sensors’ speed in each monitoring region so as to maximize the rendezvous time the sensors have at our disposal to interact. First, an idealized case where the robotic sensors are assumed to oscillate harmonically is considered, and then the more general case where these robotic agents are drifting along nonlinear ocean circulation models is studied. Similarly, the analysis of motion synchronization via local intermittent interaction starts with a two-robot model and is then extended to large-scale lattice-like spatial sensor distribution. The assumption that robots cannot overcome the dominant ambient environmental dynamics and interact only intermittently and locally significantly complicates the analysis and the control design. In such cases, the ambient geophysical dynamics that primarily drives the motion of the robotic sensors needs to be directly incorporated into the control design. To address the challenge, the dissertation presents a decentralized, intermittently activated, pairwise interacting control strategy for the robotic sensors, which under certain conditions on overall network connectivity, brings the whole system into a steady state where all robotic sensors synchronize their periodic rendezvous around configurations determined by the surrounding geophysical field. Simulation results are presented to corroborate the theoretical analysis. A small-scale experimental study validates the efficacy of a pairwise control strategy under significant noise levels and hints at extensions to large-scale mobile sensor networks.
Description
Keywords
Marine robotics, Multi-agent coordination, Nonlinear control