The sudden release of millions of gallons of oil due to an accidental spill or leak can devastate ocean and shoreline ecosystems. The April 2010 explosion on the Deepwater Horizon rig leaked 205.8 million gallons of oil into the Gulf of Mexico, causing $38 billion dollars in damages and clean-up costs. In addition to killing thousands of birds, sea turtles, and dolphins, leaked oil washed onto the salt marshes near the shore, killing plants and consequently accelerating shoreline erosion that is likely to be permanent.
Governments, corporations and researchers are continually making efforts to reduce the quantity and severity of these disasters and improve emergency response when they do occur. In the immediate aftermath of the Deepwater Horizon leak, local, state and federal agencies worked to clean up the oil using techniques that included confinement of the oil, skimming the oil from the water surface, and applying chemical dispersants to break the oil down. Some 16.5 million gallons were chemically dispersed, 6.2 million were skimmed, and 35 million were directly recovered from the wellhead, while 53.5 million gallons remained in the water or washed ashore.
One difficulty highlighted by the Deepwater Horizon response is that underwater oil plumes propagate in a complex manner which can be difficult to predict. Seeking to provide more intelligent technological support for agencies charged with clean-up efforts, Dr. Yi Guo of the Department of Electrical and Computer Engineering at Stevens Institute of Technology has won a grant from the National Science Foundation (NSF) to provide better oil plume data by deploying heterogeneous ocean robots (including wave gliders, unmanned surface vessels, and autonomous underwater vehicles) to detect and monitor the propagation of oil plumes.
“With an estimated 1.3 million metric tons of oil annually released into the sea, oil spills and discharges are ongoing problems that require organized and efficient responses,” says Dr. Michael Bruno, Dean of the Charles V. Schaefer, Jr. School of Engineering and Science. “Dr. Guo’s innovative distributed robotics will establish more effective strategies to collect and disperse oil in future spills.”
When oil is spilled in the ocean, it generally propagates through two transport mechanisms. Advection is the transport of the oil due to the motion of the water, and diffusion is the oil’s motion from areas of higher concentration to areas of lower concentration. These mechanisms can be taken into account in a general advection-diffusion model, which considers parameters like probability, time, velocity and a diffusion coefficient with spatial variability.
Dr. Guo will develop distributed multi-robot deployment algorithms for the autonomous vehicles so that their movements match such a model. The robots will be able to cooperate and maneuver themselves autonomously in order to map and monitor an underwater oil plume. They can then get the real-time sensor input of the plume concentration and other important parameters so that the source of the leak is pinpointed and the oil propagation is more accurately monitored. “The vehicles will sense and identify parameters in the oil plume propagation model and adjust their position accordingly in real time. We are using an iterative process that continually revises the tracking model, making it more accurate than any currently available.”
Presently, ocean robots field deployments are limited to individual robots, and the algorithms for autonomous vehicles have generally applied 2D models. Motivated by the gap in theoretical development and field deployment for distributed robotics, Dr. Guo aims to apply the state of the art to imperative field operations. “This research represents multifaceted advancement because 3D modeling in underwater and aerospace applications is more complex than 2D modeling, and algorithms that consider group dynamics to control a team of robots are far more intricate than those that consider only the dynamics of a single robot. In going from one robot to two, for instance, the difficulty is not simply doubled, because the algorithm must also consider their interactions with each other.”
The techniques developed in the course of their research will have long-term impacts in underwater exploration such as oceanographic survey and energy production in deep water. The results potentially benefit other environmental monitoring tasks with underlying diffusion and advection processes, such as weather event tracking and climate prediction. Furthermore, Dr. Guo’s algorithms can be extended to general high-dimensional vehicle deployment, such as aerial vehicles, enabling any robot group to work out a consensus and deploy effectively as a team.
In addition to the algorithms and simulations in development, the project also includes an experimental aspect. Brian Bingham, Dr. Guo’s collaborator at the University of Hawaii’s Field Robotics Laboratory, will test a new wave glider prototype that will serve as the leader of the team of unmanned vehicles. The wave glider has a two-part design, with a float at the surface connected to a submarine with vertically oscillating “wings”. This architecture takes advantage of the fact that wave energy is highest at the surface and weakens with greater underwater depth. As the float climbs and falls with the waves, pulling the sub up and down, the sub’s wings pivot up and down to propel the glider forward. The wave glider solely uses the endless power of the ocean’s waves for propulsion, making it an excellent candidate for leading and anchoring the robot team.
The project integrates research with education activities through robot-centric undergraduate and graduate education, robotics competition, short course and workshop development, and outreach to K-12 education. PhD candidate Shuai Li is working with Dr. Guo to relax the vehicle team’s communication topology from “all-to-all” to “neighbor-to-neighbor”, which is more reliable, efficient and robust. Because the complexity of group dynamics multiplies with each addition to the team, an algorithm which requires all-to-all communication must be revised for an expanding team. Shuai is working to necessitate only that the robots communicate with their nearest neighbors, working up to a team consensus. Instead of sending messages to every member of the group, each robot needs only to send a message to neighbors to communicate with the whole team. This establishes scalability in the algorithm, as each additional team member requires minimal extra overhead.
According to Dr. Yu-Dong Yao, Director of the Department of Electrical Engineering, “The educational component of this research gives our students a great opportunity to apply their ingenuity to an urgent international concern. The capabilities established in this research potentially allow responding agencies to more comprehensively understand where oil plumes will go and focus their efforts to respond more efficiently to a disaster. ”