Learning to write with the fluid rope trick.

Area of Science:

• Fluid dynamics
• Robotics
• Machine learning

Background:

• Direct ink writing (DIW) is a 3D printing method requiring precise nozzle control to avoid fluid instabilities.
• Fluid instabilities, like coiling or folding of viscous jets, typically disrupt printing accuracy.

Purpose of the Study:

• To investigate harnessing fluid instabilities for controlled pattern generation in 3D printing.
• To develop a deep reinforcement learning (DRL) strategy for manipulating viscous fluid streams.

Main Methods:

• A DRL agent was trained in a viscous filament simulator to learn nozzle control strategies.
• The learned strategies were experimentally validated by controlling a falling viscous jet.

Main Results:

• The DRL agent successfully derived control strategies to manipulate the fluid jet.
• Experimental results demonstrated the creation of cursive writing and abstract patterns using the learned control.

Conclusions:

• Deep reinforcement learning can effectively control fluid instabilities for precise pattern deposition.
• This approach offers a novel method for 3D printing complex structures by leveraging, rather than avoiding, fluid dynamics.