Home > Cool Tech > Most drones just snap pictures, but this one can…

Most drones just snap pictures, but this one can paint them on canvas

Connected devices might be making our homes smarter, but the means by which we decorate our work and living places hasn’t changed proportionally in the past few decades. Professor Paul Kry and his team at McGill University’s School of Computer Science may be altering that, however, courtesy of a swarm of tiny drones capable of painting walls, provided that they’re programmed with the right software.

Rather than simply painting the walls one color, Professor Kry’s work goes even further by getting drones to create pop art-style dot painting murals of some of the twentieth century’s most iconic figures. To do this, the drones use a tiny arm “holding” a sponge soaked in ink, which they then employ to dab the area to be painted. Repeat several thousand times and — hey presto! — a finished painting emerges. Already the drones have used this technique to recreate the likenesses of computer pioneer Alan Turing, actress Grace Kelly, and revolutionary political leader Che Guevara.

Related: Incredible AI app can ‘repaint’ photos to look like they were composed by famous artists

“There are a number of big empty white walls in my building on campus, in particular a long one right by my lab,” Professor Kry told Digital Trends, describing the origins of the project. “For some years I had thought about building a giant plotter, but after buying some small Crazyflie drones I had the idea of using them to draw stippled artworks: the key being that stipples are much easier to draw than continuous strokes. The flying robots have the nice advantage being able to get to hard to reach places, up high, or likewise can deal with curved walls just as easily as flat ones.”

Making the drones work requires a motion capture system to measure its position, some smart algorithms to estimate speed, and some careful tuning of parameters for feedback controllers. The drawing component is also pretty neat: involving computing a collection of dots of varying sizes so that they will produce the desired image, and a dynamic adjustment of these positions based on errors made along the way.

“Because the robot is small it can’t carry much so we use an ink soaked sponge, and the ink gets used up as we draw, producing smaller and smaller dots,” Kry said. “We estimate a model for this, and use the model in our high level path planning to draw appropriate sized dots that are nearby. There are some practical details of changing batteries every 5 or 6 minutes, as well as refilling the ink, but otherwise the robot is autonomous.”

In terms of future work, Kry told Digital Trends that the team is working to use multiple robots to paint one image simultaneously, using a tether to provide power to avoid battery changes, and even using ink cups that allow the robot to refill autonomously too.

“We expect to submit some of this follow up work to an academic conference quite soon, but there there are certainly lots of other research directions that this work can go too,” he concluded.