Thursday, August 04, 2016

On Propelling Swarms of Underwater Robots

Underwater construction, salvage, rescue, and scientific exploration can be dangerous, difficult tasks even for highly trained individuals. They can also be expensive. Enter the underwater robot. Controlled by remote or autonomously, robots explore volcanoes under the surface of the ocean, install sensors on the sea floor, search for the wreckage of missing planes like Air France Flight 447, collect military intelligence, and map the seafloor for oil and gas companies, and they do it all without threat to human life.

Their important applications have motivated researchers to explore the best ways to design underwater robots. One of the current challenges involves equipping robots to make precise movements while performing complicated tasks. If you’ve ever tried to pick up rings from the bottom of a pool, you know that controlling movement underwater is a little different than on land. New research published in the American Physical Society’s Physical Review Fluids by MIT researchers sheds light on an important, but little studied interaction that could help scientists and engineers optimize the movement of underwater robots.

Interacting jets.
Image Credit: Athanasios G. Athanassiadis
Consider a swarm of underwater robots traveling en masse to a particular location on the sea floor, at which they all spread out to complete specific tasks. What is the best way to design their propulsion systems and control their behavior? This question led MIT graduate student Athanasios Athanassiadis and his advisor Douglas Hart to sea salps.

A chain of salps
Image Credit: Henry Jager (CC BY-NC-ND 2.0)
Not the cutest sea animals, but beautiful in their own right, salps are generally 1-10cm long, translucent, gelatinous creatures. They spend the early part of their lives attached together in long chains that can span 50ft in some species and take on different shapes. They spend the later part of their lives alone. Salps are tubular creatures, pumping water in one end of body, through feeding filters, and out the other end. Not only is this how they eat, it’s how they move. Each time they expel water they are propelled forward. It turns out that salps have three important connections to efficient robot swarms.
  1. They move by jet propulsion.
  2. They maneuver together efficiently.
  3. They maneuver alone efficiently.
Pulsed jets are often used to propel underwater robots because research shows that they provide an effective means of maneuvering, and improve how well the robots travel at low speeds. In the case of the sea salp, nature has shown that pulsed jets are an efficient way to travel both as individual and in a tightly knit group.

The researchers noticed that salps pulse in sequential patterns, not all at the same time. This led them to wonder why, and if the jets from one salp might interact negatively with jets from its neighbors if they were to pulse at the same time. If so, this could affect robot design.

If you want to precisely control a robot’s motion, one solution is to give it several pulsed jets that can be fired as needed. The question is: how will the nearby jets interact if you fire them at the same time? In a swarm of robots, how will the jets from one robot interact with the jets from its neighbors? Despite these questions, there hasn’t been a lot of research on how pulsed jets interact with one another and what effect this has on thrust, the force of propulsion.

To explore this, Athanassiadis built an experiment using a fish tank and two nozzles controlled by valves. The nozzles were mounted on a rail above the tank and submerged in the tank so that, like salps and robots, they created underwater pulsed jets. Athanassiadis could easily change the spacing between the nozzles. The tank was filled with water, and the nozzles filled with water and a dye that fluoresced green under blue light. The experiment was designed to explore jet interaction in a low-speed environment, such as when an underwater robot is performing a task.

Athanassiadis trained a high-speed camera on the tank, illuminated the tank with a blue laser, and began collecting data. Each experiment lasted one second, and involved creating two pulsed jets at the same time and recording on camera how they interacted (see top image). The researchers repeated this for different spacing between the two nozzles.

In order to estimate the forces at play, the two engineers created a mathematical model that connects the strength of the thrust and the propulsion efficiency to the shape of the interaction. This gave them a foundation for analyzing the images. The result? When jets are close, they can significantly decrease the overall efficiency and thrust—by as much as 10%. When jets are farther apart, separated by 2.5 nozzle diameters or more, the jets don’t interact in a way that affects thrust and efficiency
significantly. This means that jet spacing needs to be an important consideration for underwater robot design. If you pulse multiple jets at the same time and they are very close together, you get less bang for your buck.
But the story doesn’t end there. Their model suggests that well-timed pulses from nearby jets could interact in a way that actually gives you more bang for your buck. In other words, by careful timing you can take advantage of the physics of interacting jets in a way that increases the thrust and efficiency beyond what you would get if the jets didn’t interact at all.

Like most research, arriving at these results wasn’t quite as straightforward or direct as it sounds. When reflecting on the project, Athanassiadis commented on a poster from an early talk he gave on this work. “By the time the project ended, I had scrapped all of the initial data I had taken. However, I still have a poster hanging with the old (wrong) plots on it. It's as a reminder of how the research cycle goes - science is a lot of back and forth between taking data, trying to understand what it’s saying, and then improving it until you can rationalize what’s going on and convince yourself the data is meaningful.”

Hopefully the winding, indirect path of research done today will help swarms of underwater robots follow a more precise, direct path when setting out for their tasks tomorrow.

To see the experiment in action, check out the YouTube video Vortex Rings, Salps, and Underwater Robots by Athanassiadis and Nicole Sharp of fyfluiddynamics.

--Kendra Redmond

No comments:

Post a Comment