How Robotic Fish Could Rescue Sea Life from Disasters_
Georgia State University[조지아주립대학교,미국유학설명회]
An explosion on the Deepwater Horizon Maconda oil well drilling platform in 2010 started the largest marine oil spill in United States history. Millions of barrels of oil were released into the Gulf of Mexico, polluting waters that are home to hundreds of species of fish and other wildlife.
Scientists aren’t sure what the spill’s long-term ecological impact will be, but birth defects and heart abnormalities in fish have already been documented, and bottlenose dolphins are sick and dying in larger numbers than ever before.
The lives of many fish might have been saved with the help of swimming robots, said Igor Belykh, associate professor of applied mathematics at Georgia State University.
“You see animals suffering from the oil spill, and that is our inspiration, trying to save their lives,” Belykh said. “Technology has come so far, and maybe we could have done more to help them. I’d like to see if robots could make a difference and stop something like this from happening again.”
Researchers at Georgia State and New York University have joined forces to develop a school of robotic fish with a big mission: to attract real fish in the path of disasters and steer them out of harm’s way. Fish could be saved from any number of perils, from oil spills to the spinning blades of underwater turbine systems used by power plants.
How can math help fish?
Robotic fish aren’t exactly new. The first model, dubbed “robotuna,” was created 20 years ago by researchers at Massachusetts Institute of Technology (MIT) who wanted to develop a better propulsion system for mini submarines. There have been a number of advances since then, but scientists still haven’t figured out the math that would make robots swim together like a real school of fish.
Belykh’s work could be the answer to this problem. He first started developing a concept called dynamical network theory with funding from the National Science Foundation and Georgia State’s Brains & Behavior Program, and now he’s received a $375,000 grant from the U.S. Army Research Office to study a branch of applied mathematics he believes is key to programming robotic fish to swim as a coordinated unit. The concept could also be applied to U.S. Army missions, such as formation flying of autonomous drones in hostile environments.
One robotic fish isn’t enough to attract a school of real fish. You need a whole group that can swim together. And not only that: the robots must be synchronized to move in formation, execute complicated moves and turn in sync, while keeping a certain distance from each other just like real fish. The goal is to make these robotic fish mimic the subtleties of natural movement patterns so convincingly that they gain the trust of actual schools of fish.
In mathematical terms, such a group is known as an evolving dynamical network—“evolving” because the structure of the network or group changes, Belykh explained.
“It’s a very deep mathematical problem,” he said. “Each robot needs to extend some information about each other’s location, velocity and tail frequency.”
The researchers also have to strike the perfect balance in communication. The robotic fish must communicate to perform these coordinated functions, but they shouldn’t interact too frequently. They need to exchange information only sporadically to save battery life, Belykh said.
It’s a riddle that has frustrated scientists for years. To solve it, Belykh must develop an algorithm, a set of steps for solving a mathematical problem, to control this evolving network.
Convincing real fish to play along
Belykh’s research partner, Maurizio Porfiri, a mechanical engineering professor at New York University, has always been fascinated with animal behavior. He spent years perfecting a robotic fish model, and his hard work is paying off: in recent studies, zebra fish have found his robotic fish appealing.
He separated a tank into three compartments with real fish in the center, a robotic fish on one side beating its tail on the divider and nothing in the third compartment. He measured the time the fish spent in the vicinity of the robot versus the empty compartment. The fish wanted to spend time with the robot. Some moved toward the robot and others followed it.
Fish determined whether the robot was safe based on the way it moved, Porfiri said. They will follow the robot if it accurately mimics their natural swimming motion, according to a CNN story on his robotic fish.
But the details of the robot’s appearance were important. The attraction was lost if the robot was longer, shorter or gray in color. Also, fish wouldn’t get close to the robot if it didn’t beat its tail, Porfiri explained in a recent TED Talk.
Porfiri adds color and stripe patterns to the robots that resemble the fish he’s trying to attract. He can also make the robots small or large, depending on the size of the fish he’s targeting.
The most important feature of the robotic fish is the tail, Porfiri said. He’s been able to get the tail to bend without making any noise, which would be a red flag to real fish.
“The biological problem is to find the right size, color and frequency of the tail to make real fish believe this guy can be a leader and decide to follow it,” Belykh said.
The robotic fish are programmable devices with a chip, battery, motor, Wi-Fi capability and a moving tail controlled by an iPad. Belykh’s mathematical model will be incorporated into the chips and the iPad program to guide the school’s movements. Porfiri will perform further animal studies after the robots have been programmed to swim together.
“The main thing is we’re trying to get the same type of flow between the robots and the animals,” Porfiri said.
In the meantime, Georgia State researchers are also using mathematical modeling to study brain function, locomotion and motor diseases, cardiology and cancer.
“Mathematics can be very exciting,” Belykh said.
