Robot swarms guided by human artists could paint colourful pictures

By Layal Liverpool

Small swarms of robots could help artists paint pictures. María Santos at the Georgia Institute of Technology and her colleagues designed a system that would allow an artist using a computer to select regions of a canvas to be painted in certain colours. These are then created in real-time by 12 robots that cross the canvas leaving trails of colour behind them.

“We envisioned the multi-robot system as an extension of an artist’s creative palette,” says Santos. “Because the human only controls the collective behaviour of the team, the multi-robot team can be [thought] of as an active brush for the human artist to paint with.”

Currently, the robots don’t carry real paint. Instead, the researchers tested their ability to work together using projectors that simulated coloured paint trails behind the robots on the canvas. Each of the robots can produce three primary colours – magenta, cyan and yellow – which can also be combined to produce additional hues.

Santos and her team found that the robots were able to work together effectively by varying the colours of the trails they laid down, while also considering the colours laid down by neighbouring robots.

Advertisement

“At each point in time, each robot has information about the colour concentrations desired by the human artist and can get information about the relative positions of its closest neighbours and the paints available to them,” says Santos.

She says the next step will be to develop robots that can handle real liquid paint. “This step involves not only developing the hardware necessary to manage paint, but also studying the painting release mechanism needed to achieve appropriate colour mixing.”

Vanessa Sanchez at Harvard University says another challenge with using liquid paints will be the drying time, as the robots currently run on wheels, which could result in them streaking paint across the canvas before it has dried. One way to avoid this might be to use different types of robot, such as drones, she says. “You wouldn’t have to worry about tracking of the wheels.”

Journal reference: Frontiers in Robotics and AI, DOI: 10.3389/frobt.2020.580415