Scientists have given artificial intelligence (AI) the ability to “feel” surfaces for the first time, opening a new dimension to the deployment of the technology in the real world.
Drawing on quantum science, scientists combined a photon-emitting scanning laser with a new AI model trained to differentiate between different surfaces imaged with the lasers.
The system, described in a new study published October 15 in the journal Applied opticsprojects a series of short light pulses onto a surface to “feel” it, before backscattering them. photonsor particles of light, return with a speckling noise – a type of defect that manifests itself in imaging. This is normally considered detrimental to imaging, but in this case the researchers processed the sound artifacts using AI, which allowed the system to discern the topography of the object.
“It’s a marriage between AI and quantum”, co-author of the study Daniel Tafonédoctoral student at the Stevens Institute of Technology in New Jersey, said in a statement.
The team used 31 variations of industrial sandpaper with roughness ranging from 1 to 100 micrometers in thickness – the thickest being about the width of a human hair. The researchers then set up the lidar system, which used a laser beam emitted in picosecond pulses (1 trillion picoseconds corresponds to 1 second).
Pulses of light passed through the transceivers, hit the sandpaper, then bounced back through the system for analysis by the AI. The backscattered photons came from different points on the surface and were counted using a single photon detector.
The results resulted in an average error of around 8 micrometers, but this improved to just 4 micrometers after the AI worked with several samples. This roughly corresponds to the accuracy of currently used profilometer devices.
“Interestingly, our system performed best for finer grained surfaces, such as diamond lapping films and aluminum oxide,” Tafone said in the release. These materials are often sandpapered for a specific application.
The scientists said the new method could be used for a variety of applications, including in medical settings to detect the thickness of moles that could be the precursor to skin cancer.
“Tiny differences in the roughness of the moles, too small to be seen with the human eye but measurable with our proposed quantum system, could differentiate these conditions,” Yuping Huangdirector of the Stevens Center for Quantum Science and Engineering (CQSE), said in the release. “Quantum interactions provide a wealth of information, using AI to quickly understand and process them is the logical next step.”