Artificial intelligence is, despite its sci-fi sounding appellation, a regular feature of our everyday lives.
AI is computer science dedicated to the study and development of intelligent machines and software, and solving the central problems surrounding reasoning, knowledge, planning, learning, communication, perception and even the ability to physically manipulate objects. The ultimate aim of AI research is to precisely describe human intelligence and simulate that set of processes and understandings within machines.
Since the days of Talos of Crete, the Bronze Robot of Hephaestus, various visions of what constitutes AI have come and gone.
When mathematician Alan Turing demonstrated that machines could at least simulate any conceivable form of mathematical deduction, his ideas were melded with discoveries in neurology, information theory and cybernetics, and the foundations of work on an "electronic brain" had been laid.
Now engineers at PARC, the Palo Alto Research Center, are using artificial intelligence to drive software which could help designers make informed decisions about which 3D printers and materials are best suited to make their ideas reality.
As 3D printing has the ability to manufacture objects which can't be made with traditional techniques, it's a versatile and wide-open field. It is, however, still hampered by limitations in available materials. Material specs are hard to come by, and that makes judging how thick or thin a part can be or how a given material will respond in the "real world" a crap shoot.
To solve this problem, engineers at PARC are working on an automated software program to let designers choose the best 3D printing methods and materials. This software would act as a sort of "digital manufacturing engineer" to direct designers away from materials which might be unsuitable for a particular application.
The PARC team has built a system to reference the 3D printer specifications, the materials they can accept and the properties of those materials. The software uses that information to compare a design to the geometric and material parameters used for printing and then simulate a print run. The "virtual print" is then used by the PARC software to detect where an object might fail and to then suggest manufacturing changes.
Once made available to the general public, engineers would have a tool to give them more realistic and useful designs.
PARC CEO Stephen Hoover says the goal of the project is to build programs to enable non-experts to "think their way through a design space" before outputting a product or object.
Hoover says while engineers with tons of experience can eliminate whole sets of avenues which might cause production problems, novices lack that sort of savvy.
Tolga Kurtoglu, chief of PARC's design and digital manufacturing program, says many of the types of constraints that can cause problems for novice designers can be solved with software.
As each material has minimum thickness requirements, designs that fail to comply with this constraint can be printed but are more likely to encounter problems like distortion or warping during output. He says the initial system his group has built includes libraries to help their tool assess the feasibility of printing a given design on a specific type of printer. Geometric features of the design are compared with the parameters dictated by the printer and material, and then the software simulates layer-by-layer production of the proposed object to expose problematic areas.