January 4, 2008

Dolphins Play with Robotic Seaplane

It's a bird, it's a plane -- no, it's the Flying Fish, a new unmanned
seaplane developed at the University of Michigan that has demonstrated
the ability to take off, fly and land autonomously in moderate seas
some 6 feet high.

Researchers at U-M's Aerospace Engineering Department and its Marine Hydrodynamics Laboratories who designed and built the new unmanned aerial vehicle (UAV), which they named the Flying Fish, think it is the first UAV that can take off and land on water all by itself.

The electrically powered Flying Fish was able to take off, fly and land
autonomously 22 times during two days of sea trials held late last year
off the coast of Monterey, Calif.

It also became probably the world's most unusual dolphin toy. Remarkably, when the UAV was landing in the sea, scientists on the research vessel John Martin monitoring the tests saw dolphins swim over to investigate.

To the researchers, it appeared that the dolphins thought the UAV was a
pelican diving for fish, said Ella Atkins, a researcher from U-M's
Aerospace Engineering Department.

Finding the craft wasn't a bird that had spotted tasty fish swimming
below, the pod of three dolphins began playing with it, repeatedly
swimming in formation with the little seaplane and diving underneath on
one side to surface again on the other.


It's not surprising the playful dolphins initially were fooled. The
Flying Fish is about the size and weight of a large seabird.

"We studied seabirds seriously," said Guy Meadows, director of the U-M
Hydrodynamics Laboratories. He had the idea for the unmanned seaplane
while watching flying fish pop up from the sea, soar over the waves,
and drop down again.

Seabirds are "all about the same size -- about 20 pounds with a 2-meter
wingspan," said Meadows. "It turns out that, aerodynamically speaking,
that's a sweet spot to be flying close to the water. Our plane is about
the size of a large pelican."

That didn’t mean the Flying Fish was shaped like a seabird. "It had a seaplane shape with dual pontoons," said Atkins.

The UAV's body and catamaran-like twin pontoons make use of Meadows'
speedboat-hull-design experience to direct sea spray away from its
wings and help the craft "get up" quickly on the water when it begins
its take-off run, she explained.

U-M's Flying Fish isn't the first unmanned seaplane. That honor
probably belongs to the 300-pound Sea Scout, developed by Oregon Iron
Works. But the Oregon Iron Works UAV doesn’t appear yet to have
demonstrated the ability to take off autonomously on water.

Autonomous sea take-off crucial

However, the ability to do so was crucial for the electrically powered
Flying Fish, which the U-M researchers designed and built in just eight
months in 2007 to meet the Phase 1 goals of the Defense Advanced
Research Project Agency's (DARPA's) "persistent ocean surveillance"

"The difficulty of taking off on the ocean is that you either have to
predict the oscillations of the ocean surface, or ignore them," said

But had the pitch control of the UAV been designed automatically to
respond to the vertical oscillations of the waves while taking off,
response-time delay would have led to the controls over-correcting for
each upward and downward wave movement.

This "induced oscillation" would have made it impossible for the aircraft to take off, said Atkins.

Instead, the researchers chose to solve the problem by effectively
ignoring wave oscillations and letting the Flying Fish just plough
through the waves on its short take-off run.

After performing take-offs controlled by a pilot using a remote-control
radio transmitter, they fed recordings of the pilot's radio-signal
inputs into the onboard processing unit that is used to control the
UAV's flight surfaces during autonomous flight. Those recordings
controlled each autonomous take-off.

Need to stay within a "watch circle"

A major initial goal of the demonstration, which was performed in front
of DARPA officials, was to show that the UAV could monitor and remain
within a defined "watch circle."

The idea was for the Flying Fish to drift at the center of the circle
until its onboard GPS-based navigation system told the craft it was
floating too far out of position. That triggered its autonomous
take-off sequence, the little seaplane ploughing through the waves for
just 10 meters before lifting off, flying and reaching a point where
its GPS unit told the UAV it should land again.

The unmanned aircraft had to demonstrate that it could fly across the
watch circle area completely autonomously: taking off, climbing,
cruising and descending. The craft also had to acquire data all the
while, through the onboard inertial gyro sensors it used to measure
roll and pitch and the pressure sensors it used to measure airspeed.

Landing essentially involved a shallow descent.

"When it impacts the water, it goes, 'Oh, there's the water,'" said
Atkins. "The boat has very well-designed pontoons. Because it doesn’t
have a flat bottom, it cuts into the water like a diver, as opposed to

The tests were successful. "The vehicle was stable and handled very well," said Atkins.

GPS navigation

GPS was suitably accurate
as a navigation aid because, while the up-and-down lateral movement of
the ocean waves represented a position-fixing problem, the aircraft's
navigation system wasn't constrained by having to land on a defined,
narrow runway -- the Flying Fish could take off and land anywhere on
the sea within a given area.

Following the successful tests, the Flying Fish is now back in the
shop, the U-M researchers planning to outfit it with solar power and
more sensors.

"One of the plans for the next generation is to do energy harvesting with solar cells," said Atkins.

While solar-powered high-altitude aircraft designed to take off from land and stay aloft for years at a time
"have to charge energy quicker than they use it," the Flying Fish would
have no such constraint. Every time it needed to recharge its
batteries, it could simply land on the ocean and float for a while.