MAE-led team advancing research in cyber-physical oceanography

A collaborative team of researchers from the NC State and the University of Michigan, led by NC State MAE Associate Professor Chris Vermillion, has been selected for a $1.2 million, three-year NSF project to create and validate adaptive mission planning strategies for cyber-physical oceanographic modeling.

The team is focused on planning persistent missions for a network of observing instruments, including renewably powered data collectors. Persistent missions, as opposed to one-time deployments performed over just a few days at costs of hundreds of thousands of dollars, are critical for obtaining the detailed observations needed to characterize climate-related events and trends that currently are only sparsely modeled. One focus of this project is the complicated currents offshore of North Carolina, where the Mid-Atlantic Bight, South Atlantic Bight, and Gulf Stream meet. There, the renewably powered instrument will be a SeaTrac SP-48 solar-powered autonomous surface vessel. The mission planning methods developed there will be applicable across a broad variety of data collecting networks.

The research team combines expertise in control, energy harvesting, robotics, and physical oceanography. Vermillion studies control and energy harvesting. Dimitra Panagou (University of Michigan Robotics) focuses on robotics and control, and Ruoying He (NC State Marine, Earth, and Atmospheric Sciences) works in physical oceanography.

Amid global climate change, obtaining a detailed understanding of ocean processes through continual, high-resolution monitoring has never been more important. This is especially true along the east coast of the U.S., where large variations in Gulf Stream position, extreme wintertime wind stress and water density changes, accelerated warming of continental shelf water, and sea level rise north of Cape Hatteras have been documented in recent years. As Co-PI He notes, “These recent trends are warnings of potential larger shifts in atmospheric and oceanic conditions, yet how these changes on the continental shelf impact the open ocean are unknown. We need persistent, adaptive data collection so that we can refine our ocean models.”

In response to this need, a number of companies have developed wind-, wave-, and solar-powered vessels for unmanned observations. These technologies are mostly at an early stage of development, and their deployments are typically pre-planned, solo missions of limited duration. The innovation of this project is to use these renewably powered vessels as “hosts” that can recharge “satellite” vehicles, which can then perform longer-term, more detailed data collection. As Vermillion puts it, “The robotic hardware exists for performing persistent missions, but the mission planning approaches remain largely stuck in the past.” By forming a multi-disciplinary team at the intersection of control, robotics, energy harvesting, and oceanography, the team aims to create adaptive mission planning approaches to achieve longer-term, crucial oceanographic missions.

Besides having immediate relevance for physical oceanography, this research addresses several fundamental and cross-cutting Cyber-Physical Systems challenges. Specifically, the team will focus on managing a critical tradeoff between the cyber quantity of information and physical quantity of energy. Unlike short-term missions where the available energy is finite and is a hard constraint within an optimal control formulation, the available energy for a renewably powered persistent mission is nearly infinite over time but uncertain at any instant. Thus, the mission planner must make an accurate estimate of the informational value (i.e., oceanographic data that can be acquired) of stored energy at the end of any finite prediction horizon.

The team is approaching this challenge by characterizing information based on a scientifically tailored dynamic coverage model, using a predictive control framework where the estimated informational value of energy informs a terminal incentive in the finite-horizon optimization. To address the challenge of deploying and recovering satellite agents (UAVs or undersea gliders) from the host in an information-optimal manner under the presence of energy constraints, the team is turning to energy-aware coverage algorithms under development by University of Michigan Panagou’s lab. As Panagou notes, “This project serves as a tremendous advancement of existing energy-aware coverage algorithms, which rely on a priori partitioning of the mission domain and have not yet addressed energy-aware coordination between an energy-limited agent and its mobile host.”

In addition to its direct impact on Cyber-Physical Systems and physical oceanography, this research will further expand the portfolio of renewably powered robotic systems being explored within Vermillion’s lab and collaborating groups. This portfolio includes robotic sailboats (and corresponding control software), underwater kites for recharging autonomous underwater vehicles, and (with this new award) solar-powered autonomous surface vessels. As Vermillion puts it, “Building a research portfolio in sustainable robotic systems is an extremely rewarding endeavor that allows my colleagues and me to work at the intersection of the critical and timely topics of control systems and energy, while also having a lot of fun!”