McGill researchers teaching drones to paint murals

Drones can do a lot of things, good and bad. But using drones to paint? That’s a good thing.

There are a few jobs, such as bartender or psychologist, that you would have thought would be safe from being made obsolete in this age of automation, and that “mural painter” would probably remain on that list of safe jobs.

Not so fast, says a team of McGill Computer Science researchers.

In an article called “Stippling with Aerial Robots”, published via the European Association for Computer Graphics, or Eurographics, the researchers outline their method of using a quadrotor flying robot to paint dots on a canvas, controlled by a combination of motion capture technology and an algorithm that calculates a variety of factors to have the drone reproduce an image as closely as possible.

With funding from the Natural Sciences and Engineering Research Council of Canada, the Fonds de recherche du Québec – Nature et technologies, and the Canada Foundation for Innovation, Paul Kry, Associate Professor at McGill University’s School of Computer Science, assembled a team of students to help develop the project.

The idea for the project originated with the idea that it might be nice to put some murals of famous computer scientists, like Alan Turing, in the stairwells at McGill’s School of Computer Science.
Mounting a tiny ink-dabbed sponge at the end of a rudimentary paintbrush-like extension, the small drone tries to reproduce a dot-style drawing, like the type produced for colour comic books in the 1960s or Georges Seurat paintings.

A motion-capture system perched above the canvas watches a couple of retroflective markers on the drone as it flies, keeping track of where the robot is in space, and then sends it a command telling it where to apply the next dot.

Each drawing is composed of several hundred to a few thousand black dots varying in size depending on the amount of ink left in the drone’s paintbrush and the pressure it applies to the canvas.

While this stippling technique might work well in hallways, it will be a while before drones take your outdoor sign painting job, owing to the fact that even the slightest breeze or disturbance can knock the featherweight drone off balance, messing up its accuracy.

Master’s student Brendan Galea describes having to switch to working the overnight shift because other human activity like opening and closing doors threw the robot off its game.

The picture of Grace Kelly that the drone attempted to paint has an unintentionally symbolic tear running from one of its eyes, created by an overloaded brush the drone dabbed on to the paper.

While the drone won’t likely be stealing jobs any time soon, Kry has ambitions to introduce it into the wider world.

“There’s this wonderful mural festival in Montreal, and we have giant surfaces in the city that end up getting amazing artwork on them,” he says. “If we had a particularly calm day, it would be wonderful to try to do something on a larger scale like that.”

The drone can apply about 70 dots before losing its battery charge, forcing it to land so that a student can hook it up to a new battery.

Kry and his students’ article won a “best paper” prize in May at an international symposium in Lisbon on computational aesthetics in graphics and imaging.