Students Learn Control with Quadcopters

2013-01-16

Suspended in air, the unmanned quadcopter hovered, unsure how to proceed through two large obstacles: a “sphere” (actually, a trash can), and a “pillar” (a tall lab stool).

A student behind the controls returned to the computer, modified the aircraft’s settings for repelling from objects, and tried again. The quadcopter took off, hesitated a bit when approaching the obstacles, then passed between them cautiously. Success!

Another drill involved a quadcopter tethered with a string holding a magnet on the other end. A second magnet was placed strategically on the floor. With a few passes, the vehicle sniffed but missed catching the second magnet. The student adjusted a grid. After a few more tries, the quadcopter hit the spot, the magnets connected, and the target was lifted from the floor.

“We are the champions!” the student cried in triumph.

Graduate student Mike Dorothy smiled and shook his head when asked whether this was a competition. “No,” he said, “It’s just for bragging rights.”

As playful as it appeared, the lab for AE483, Aerospace Decision Algorithms, was an innovative means this past fall to teach control, dynamics, modeling and motion planning to Aerospace Engineering at Illinois seniors. The course had a significant laboratory component, requiring students to implement and test algorithms for automatic control of quadcopters (similar to helicopters). The quadcopters, equipped with a circuit board that Laboratory Coordinator Dan Block fashioned and sensors the students built, made possible a dynamic learning experience.

“These units (the quadcopters) allow us to get down to the math to model the system,” said Block, who directs the Control System Laboratory for Engineering at Illinois.

Reflective tape balls placed on top the quadcopters and on the obstacles appeared as bright, white lights to 18 infrared cameras hung throughout the lab. Those images fed back to computers, supplying the vehicles’ positions and orientations, so students could track the aircrafts in relation to their grids and the obstacles, and make adjustments for flights.

Early in the fall, the students had used remote control vehicles bought from a toy manufacturer. But the toys’ software wasn’t easily changeable, Block said, and so control had to be accomplished from one laptop. Another downfall was that wireless devices like iPhones that students would bring into the lab would interfere with the signal for the toys. The quadcopters solved the interference issue and “are much more conducive for this class,” Block said.

AE Graduate assistant Miles Johnson said the students were divided into teams to determine controls and run through the drills. The teams used varying means to achieve the desired results. “They work experimentally to find what solution works best,” he said.

Johnson was a big factor in the lab’s success, said his advisor, AE Assistant Prof. Tim Bretl. “(Miles) took complete responsibility for the lab; among other things, writing a manual from scratch.”

Bretl was pleased with the way the course turned out and said it gave many of the students their first chance to actually fly something. “(The students) didn’t need new theory; they needed an opportunity to apply what they had already learned to make a real aerospace system work,” he said.