Autonomous drone brain being tested at University of Manitoba

Autonomous Drone Brain University of Manitoba

This post originally appeared on http://www.metronews.ca/news/winnipeg/2016/08/17/autonomous-drone-tech-being-tested-in-winnipeg.html.

[Editor’s Note: University of Manitoba professor, Witold Kinsner, sees drones flying longer distances with much heavier loads.]

‘Think of a big drone carrying crates with the components needed for a make-shift hospital shelter, tthat has to be airborne for days on end.

Here’s something that might be over your head: near-sentient drones shipping critical goods to remote Northern communities.

University of Manitoba professor Witold Kinsner said it’s not only possible, but may be necessary, which is why he and a visiting intern from China are currently working on bringing the advanced auto-pilot necessary for the task to life.

“Specifically in Manitoba, our roads are melting, delivery will have to be provided one way or another—an airplane cannot land on a dime, so it is necessary to develop such technologies,” he said. “Amazon is working on delivering small parcels… this has to be a much, much heavier load, flying much longer distances.”

Dr. Witold Kinsner.

BRAEDEN JONES/ METRO

Dr. Witold Kinsner.

Think of a big drone carrying crates with the components needed for a make-shift hospital shelter, that has to be airborne for days on end.

The weighty load and long flight time pose several practical and engineering problems: the drone cannot rely on user input or GPS signals, which are spotty in the north, and the autopilot has to be extremely precise to avoid wasting energy and cutting a flight short.

“That’s where research now kicks in,” Kinsner said.

Cue Chen Qiu, a student from Wuhan University in China, who’s working with Kinsner on the advanced mathematics-based autopilot as part of a 12-week internship through Mitacs Globalink.

She said the inability to anticipate what variables might affect the drone mid-flight are the biggest problem when it’s flying out of contact-range.

“As we all know, when the drone is flying it may face different challenges, like a sudden change of wind speed or direction, or a sudden change in temperature, or maybe it rains,” she said. “To achieve this long distance flying, we should design a controller that can self-adjust when it meets these challenges.”

That’s been her task this summer, and it’s one that she’s made some progress on during two-dimensional testing and computer simulations.

So far in her research, she’s narrowed down her goals for the advanced controller to first minimize erroneous readings, secondly give it the “fastest possible response,” and thirdly to draw “minimal power” when making adjustments to stay on course.

“Overshoot,” or over-correcting, could result in a crash and therefore lost cargo, or worse if the drone is over populated land.

Qiu has some early ideas about how a pre-programmed controller with fixed parameters can work to keep a drone on course, but looking forward she sees limitations with that system.

“If I change the parameters, say, if the mass changed, or external force changed, should I now be tuning those parameters again?” she said.

The answer is yes, but without user-input, that drone won’t know what to do, so she plans to add an “improved control system” to “change with the change,” something like a “neural network,” that would allow the drone to learn… on the fly.

Kinsner said a cognitive system or “low-level artificial consciousness” like Qiu describes, would enable drones to make optimal adjustments; But in Qiu’s short 12-week internship, they’ve just scraped the surface of applying these new techniques to their research.

Chen Qiu.

BRAEDEN JONES/ METRO

Chen Qiu.

Which is why, with Qiu’s enthusiasm for the project, he anticipates that their collaboration will extend “beyond her time here in Winnipeg.”

Mitacs

In a statement, Alejandro Adem, CEO of Mitacs, which facilitates knowledge transfer by matching students with Canadian researchers, said building “international research partnerships” like the one Qiu and Kinsner have struck bolsters Canada’s global competitiveness.

Adem explained the economic impact of foreign researchers choosing Canada for further study is estimated to be more than $8 billion annually.

Since 2009, Mitacs, which is funded locally by Research Manitoba and the Government of Canada, has matched 2,500 international students with Canadian researchers.

A 2015 survey of former Mitacs interns found 65 per cent intend to develop, or have already developed, collaborations with Canadian researchers.

“It’s essential to promote knowledge transfer through international connections and collaborative partnerships,” Adem said.

In the future, the one Qiu and Kinsner have may be the answer to getting goods north, and potentially elsewhere.

FacebookTwitterGoogle+Share

Leave a Reply