Integrated vision improves 3D printing functionality
A 3D printer integrating machine vision has been developed by researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). The researchers say the system offers higher accuracy and convenience over traditional 3D printing.
The MultiFab system uses machine vision 3D scanning techniques to self-calibrate and self-correct during the printing process, with the system’s feedback loop detecting errors from 3D scans and generating correction masks for each layer of the design.
The vision functionality also gives users the ability to embed complex components, such as circuits and sensors, and print directly onto the body of those parts. This means the system can produce a finished product; the researchers have been able to print a lens directly on top of an LED, for instance. Another example is to print a smart phone case around the phone itself, using the image data of the phone to print the case.
MultiFab has a resolution of 40µm and can print 10 different materials at once using 3D scanning techniques. The CSAIL team presented a paper on the MultiFab system at the Siggraph computer-graphics conference earlier in the month.
‘The platform opens up new possibilities for manufacturing, giving researchers and hobbyists alike the power to create objects that have previously been difficult or even impossible to print,’ said Javier Ramos, a research engineer at CSAIL who co-authored the paper with members of professor Wojciech Matusik’s Computational Fabrication Group.
The researchers envision an array of applications in consumer electronics, microsensing, medical imaging, and telecommunications, among other things. They plan to also experiment with embedding motors and actuators that would make it possible to 3D print more advanced electronics, including robots.
MultiFab mixes microscopic droplets of photopolymers together that are then sent through inkjet printheads. The computationally intensive process, which involves analysing dozens of gigabytes of visual data, can be much more easily scaled to larger objects and multiple materials.
Ramos says that he could imagine printers like MultiFab being used by researchers, manufacturers, and consumers. ‘Picture someone who sells electric wine openers, but doesn’t have $7,000 to buy a printer like this. In the future, they could walk into a FedEx with a design and print out batches of their finished product at a reasonable price,’ he said. ‘For me, a practical use like that would be the ultimate dream.’