Engineers at Washington University in St. Louis have built a tiny optical sensor that uses artificial intelligence to measure multiple simultaneous physical changes from a single device.
Conventional instruments measure a single type of change. The WashU team built a miniature light-trapping device known as an optical microresonator.
Inside the microresonator, light travels along numerous distinct paths. Conditions change around the device. Each specific light pattern reacts in its own way. Unique optical barcodes record the alterations.
Machine-learning algorithms decode these complex light patterns.
“What excites us most is that a single, tiny optical sensor can perform functions that typically require multiple sensors,” said Lan Yang, the Edwin H. & Florence G. Skinner Professor of Electrical & Systems Engineering at the McKelvey School of Engineering.
Yang led the research alongside the group of Chenyang Lu, the Fullgraf Professor of Computer Science & Engineering at WashU. Their findings were published in Laser & Photonics Reviews.
By using multiple sensing channels and analyzing how different optical modes respond to changes, we can extract much richer information from one device. Yang added
“The sensor captures a wealth of information, but the challenge is making sense of it,” Lu said. “Machine learning helps us decode these complex light patterns and turn them into useful measurements, enabling a single sensor to detect and distinguish multiple physical changes.”
The engineering team tested a miniature, keyboard-like platform featuring nine distinct touch locations. They used a single microresonator to pinpoint precisely where the surface was pressed and how far it moved.
The system maintained its effectiveness even when optical measurements grew noisier. First author Jie Liao, a postdoctoral research associate in Yang’s group, said the strategy could eventually support compact devices.
“With further development, this strategy could support compact, intelligent devices for robotic touch, environmental monitoring and biomedical sensing,” Liao said. “For example, future health-monitoring technologies could benefit from measuring several related physiological changes together and provide a more complete picture than any single measurement alone.”
Co-author Hangyue Li said that photonic sensors could find a home in cybersecurity. Li is a doctoral student in Lu’s group.
Photonic sensors add new layers of user authentication to everyday electronics like smartphones. “People are interested in adding another degree of authentication,” Li said. “The pattern you type in—including how much pressure you apply on different numbers—that could increase the safety.”
Founded in 1853, Washington University in St. Louis held its first classes in downtown St. Louis in 1854. The institution relocated to its current campus northwest of Forest Park in 1905, after its newly constructed buildings were temporarily used to accommodate the 1904 Summer Olympics and St. Louis World’s Fair.
The study received support in part from the AI for Health Institute at WashU.