Professor L. Mahadevan, Lola England de Valpine Professor of Applied Mathematics at the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), and Professor of Organismic and Evolutionary Biology, and of Physics in the Faculty of Arts and Sciences (FAS), led a team of researchers that trained a machine to paint like Jackson Pollock.
The team asked if 3D-printing could apply Pollock's distinctive techniques to quicky and accurately print complex shapes. The team combined physics and machine learning to develop a new 3D-printing technique that quickly creates complex physical patterns. It does this by mimicing the same natural fluid instability that Pollock used.
The study published in Soft Matter addresses the slowness of 3D and 4D printing. Though revolutionary, 3D and 4D printing is bound by the law of physics, in that liquid inks falling from a height become unstable, folding and coiling in on themselves. 3D and 4D printers place the print nozzle millimeters from the surface to help with eliminating the instability of the liquid.
Mahadevan and team turned to the physics to help. Combining the physics of coiling with deep reinforcement learning, the machine could learn from its mistakes and with each trial, become more accurate. The team printed a series of complex shapes and even decorated a cookie with chocolate syrup.
“Harnessing physical processes for functional outcomes is both a hallmark of intelligent behavior, and at the heart of engineering design. This little example suggests, once again, that understanding the evolution of the first might help us be better at the second,” said Mahadevan.
Image: A 3D-printed cursive "Cambridge" printed using reinforcement learning (Soft Math Lab/Harvard SEAS).