Software that lets a chopper learn aerial tricks by watching another
Washington, September 6 : Just like birds learn to fly by watching other birds,
helicopters can also learn complex aerial tricks and manoeuvres by looking at each other,
thanks to a piece of software written by Stanford University computer scientists.
The researchers have revealed that their software could learn and flawlessly replicate
more than 20 years of radio-controlled helicopter expertise in just 10 minutes.
Though the software is not specific to choppers, the researchers have revealed that
they have been approached by private companies to make helicopters choppers that can
monitor humanitarian disasters, track wildfires or locate land mines.
"The goal was to take an off-the-shelf helicopter and write a program to fly it as good
as an expert," Discovery News quoted Adam Coates, one of the scientist involved in the
project, as saying.
"We are now more accurate and consistent than an expert human-piloted helicopter,"
added Pieter Abbeel, another Stanford scientist involved with the project.
The researchers have revealed that they decided to work with helicopters because they
are inherently unstable.
"The dynamics of helicopter flight are incredibly complicated; blades are flexing, air
is churning, etc. It's simply too complex for us to map out mathematically," said
Coates.
The researchers have revealed that instead of directly teaching a helicopter, their
software lets the computers teach themselves with the aid of data gathered through a host
of sensors and equipment.
The helicopters themselves are equipped with accelerometers, gyroscopes and
magnetometers that monitor a helicopter's speed, acceleration, direction and a host of
other variables. Ground-based video and positioning instruments gather more data about the
helicopter's performance.
The data collected from ground and air are later fed into a computer for analysis.
According to the research group, a larger helicopter could carry the entire instrument
and analysis package.
While the cameras rolled and instruments recorded, Garett Oku, an expert
radio-controlled helicopter pilot, sent one helicopter into a series of flips, rolls,
twists and other complex maneuvers.
The researchers say that their software can even let a helicopter to learn "tic toc", a
difficult aerial trick where the helicopter's nose points straight up and it swings side
to side like a pendulum.
During the tests conducted to determine the efficacy of the software, a helicopter flew
the same 10-minute routine several times.
Ten minutes after the final demonstration flight, the computer had turned the
helicopter's 20 years of training and experience into data that another helicopter then
used to create flawless flights, one after another.
"For an expert helicopter pilot to fly the same exact path over and over is very
impressive. Some of them spend years trying to do it," said Coates.
Professor Eric Feron of Georgia Tech, who worked on autonomous helicopters several
years ago when he was at the Massachusetts Institute of Technology (MIT), says that the
Stanford team has pushed the limits of autonomous helicopter flight and computer
programming.
"What I'm most impressed with is the learning part, the ability of the algorithm to
learn and to fly and then to reproduce that in another aircraft. No one had done that
before," he said.
The computer algorithm can only copy the moves of a human pilot, and it cannot think
independently or creatively.
However, still, both Coates and Abbeel believe that that is certainly a possibility for
the future.
--ANI
