Artificial intelligence-based applied sciences are rapidly evolving to perceive, communicate, compute, and innovate. One area where they consistently fall short is measuring or “feeling” surfaces – a purely mechanical task that requires little emotional intelligence.
According to Stevens physics professor Yong Meng Sua, AI has nearly attained the capability of visual perception through significant breakthroughs in laptop vision and object recognition capabilities. Despite lacking sentience, the algorithm has surprisingly acquired a human-like ability to detect texture, permitting it to distinguish between rough, thick newspaper sheets and smooth, glossy magazine pages.
Till now, that’s. In a breakthrough at Stevens’ Center for Quantum Science and Engineering, researchers have successfully developed an innovative approach to imbuing artificial intelligence with emotional capabilities.
Súa collaborates with Dr. Yuping Huang, Director of the Center for Quantum Systems Excellence (CQSE), alongside doctoral students Daniel Tafone and Luke McEvoy, a 2022 Master’s graduate. In 2023, researchers developed a cutting-edge quantum laboratory setup that combined a photon-emitting scanning laser with advanced AI algorithms trained to distinguish between various surface types based on images taken with these lasers.
According to Tafone, this development represents a groundbreaking fusion of artificial intelligence and quantum technology.
Within the pages of a recent journal volume, researchers revealed groundbreaking insights into their innovative system. 63, No. Thirty people gathered around a specially designed beam of sunlight, which was pulsed in brief, intense blasts onto the floor, allowing them to “feelingly” experience its warmth and texture. Incoherent light, scattered by the target, returns to the detector, bearing speckle noise – an inherent imperfection inherent in optical imaging.
While speckle noise is often viewed as a hindrance to producing sharp, accurate images. Notwithstanding the Stevens group’s system employs a distinctive approach: it identifies and processes noise artefacts through an AI trained with meticulous precision to perceive them as valuable data. This capability allows the system to accurately pinpoint the terrain features of the object.
Taffe explains that his team employs a technique utilizing fluctuations in photon intensity across distinct lighting conditions to optimize floor design.
Staff utilized 31 industrial-grade sandpaper specimens, each featuring distinct surface textures varying in roughness, ranging from 1 to 100 microns in thickness, for the purposes of experimentation. With a mean human hair measuring approximately 100 microns in thickness for comparison, mode-locked lasers produced gentle pulses targeted at the samples.
Pulses transmitted through transceivers interacted with sandpaper-like surfaces before being redirected back into the system for analysis by a study model employed by the staff.
The group’s methodology initially exhibited an average root-mean-square error (RMSE) of approximately 8 microns during early assessments; however, subsequent work with various samples and averaging outcomes across them resulted in a substantial accuracy improvement, ultimately falling within 4 microns – comparable to the precision achieved by current industrial profilometer units.
According to Tafone, the system performed most efficiently with high-resolution surfaces, specifically diamond-lapped film and aluminum oxide.
The newly developed methodology has great potential to benefit numerous applications, according to him.
Despite their best efforts, human examiners occasionally misidentify seemingly benign lesions as potentially life-threatening melanomas when attempting to detect skin cancers.
“With precision down to the molecular level, even minute fluctuations in surface texture can distinguish between these conditions,” Huang clarifies.
Quantum interactions generate vast amounts of complex data, where leveraging artificial intelligence to quickly analyze and process this information is the next natural progression.
Manufacturers’ ability to produce high-quality parts often relies on incredibly precise measurements, where minuscule variations can mean the difference between a flawless component and one harboring a minute flaw that could eventually lead to catastrophic mechanical failure?
As expertise in LiDAR technology has been thoroughly applied in areas such as autonomous vehicles, smartphones, and robots, Huang concludes that our approach enhances their capabilities by incorporating ground property measurement on a minute scale.
