An intelligent optical sensors that process and compresses data inside the sensor itself, enabling faster surgical decisions, lower energy use, and real-time tissue analysis without relying on cloud computing.
Researchers at Texas A&M University have developed a new class of intelligent optical sensors that can analyse data directly inside the sensing hardware, potentially transforming robotic surgery, space exploration, and edge AI systems. Unlike conventional optical sensors that only capture raw data for later processing, the new devices can compress and interpret information in real time.
The technology, called electrochromic hyperspectral embedding (ECHSE), shifts computing intelligence from external processors into the sensor itself. This reduces the need for high-bandwidth data transfer and large AI accelerators while improving speed and energy efficiency. The framework was recently published in the journal Nature Sensors.
The research team says the sensors could help surgical robots distinguish between healthy tissue and tumors during operations without sending imaging data to remote systems for analysis. By performing spectral interpretation directly at the sensor level, the system can support near-instant surgical decisions, reducing latency in critical procedures.
Beyond healthcare, the researchers also see applications in autonomous space systems. Current lunar exploration missions rely heavily on transmitting imaging data back to Earth for mineral analysis. The new intelligent sensors could allow astronauts or rovers to analyse rock composition locally in real time, helping detect water ice or rare minerals more efficiently.
The system is designed to support adaptive sensing, where the hardware dynamically tracks only the most useful information instead of processing entire datasets. According to the researchers, this could lead to ultracompact machine vision systems with lower power requirements for industrial automation, robotics, and wearable electronics.
The project combines expertise from materials science, photodetectors, machine learning, and hardware engineering. The team is now exploring scalable manufacturing approaches aimed at making the sensors compatible with commercial production and industrial deployment.
