Researchers within the Charles V. Schaefer School of Engineering and Sciences have developed an innovative AI-powered quantum sensing method that enables artificial intelligence to discern surface textures. This breakthrough, achieved by a team at the Center for Quantum Science and Engineering (CQSE), marks a step toward giving AI a sense of touch, a complex capability beyond current vision-based systems.
AI has demonstrated proficiency in seeing and interpreting visual information through computer vision and object recognition for years. However, replicating the human sense of touch has yet to be discovered. “AI has more or less acquired the sense of sight,” explains Yong Meng Sua, a Research Assistant Professor of Physics, “but it has not developed a human-like sense of touch to differentiate, for instance, the roughness of newspaper paper versus the glossiness of magazine paper.”
The team, led by CQSE Director and Physics professor Yuping Huang, has overcome this limitation using a unique blend of quantum optics and artificial intelligence. Working alongside doctoral candidate Daniel Tafone and Luke McEvoy ’22 M.S. ’23, they crafted a sophisticated setup that uses a photon-emitting laser and an AI model capable of analyzing surface textures with remarkable accuracy.
Their method, recently published in Applied Optics, involves shooting a rapid series of light pulses at a surface, which then reflect to the sensor, carrying distinctive patterns of “speckle noise.” Traditionally seen as a hindrance to clear imaging, these noise patterns provide a wealth of information that reveals fine details about the surface’s topography when processed by the team’s AI model. The method allowed the researchers to analyze variations in texture as subtle as a few micrometers, even down to surfaces smoother than the width of a human hair.
“We use the variation in photon counts across different points of illumination on the surface,” Tafone said. This approach allowed them to accurately analyze industrial materials, such as sandpapers with textures ranging from 1 to 100 micrometers, achieving a root-mean-square error (RMSE) of only 4 micrometers. This level of accuracy is comparable to state-of-the-art profilometer devices, which are the current industry standard.
The potential applications for this new capability are significant, ranging from healthcare diagnostics to manufacturing quality control. “In medicine, this could aid in skin cancer detection by identifying microscopic differences in skin roughness that are too subtle for human eyes to detect,” Huang explains. Such distinctions assist doctors in identifying cancerous growths more effectively, avoiding misdiagnoses of benign conditions.
Manufacturing industries could also benefit. Detecting tiny material flaws could prevent defects that lead to more significant mechanical failures, improving the safety and durability of components used in everything from aerospace to automotive production.
The quantum-AI fusion behind this touch-sensitive technology may have implications for widely used sensing technologies like LiDAR, which operates in autonomous vehicles, robots, and smartphones. By adding a “sense of touch” to these devices, the team envisions improved detection capabilities for surface properties at minuscule scales, opening doors for applications in both consumer technology and critical industrial functions.
As Stevens continues to push the boundaries of AI and quantum technology, this development exemplifies the institution’s drive toward practical solutions that address real-world challenges. The combination of interdisciplinary studies and innovations poises Stevens to continue pioneering AI development.
