Domestic researchers have developed next-generation intelligent sensing technology capable of efficiently processing large-scale data generated from sensors. By reducing bottlenecks during data transmission, the potential for utilizing real-time artificial intelligence (AI) systems, such as autonomous driving and robots, is anticipated.

Ajou University announced on the 22nd that a research team led by Professors Seo Hyung-tak and Kumar Mohit of the Department of Advanced Materials Engineering and the Department of Energy Systems Engineering at the Graduate School has developed 'Electric Prism (E-PRISM)' technology.

E-PRISM is a technology that compresses and processes data from multiple sensors into a single signal. It is a structure that electrically implements the principle of an optical prism focusing light into a single beam.

The research team utilized a zinc oxide (ZnO)-based memristor device. A memristor is a next-generation semiconductor with the characteristic of remembering current flow information.

It is a method that arranges 10 memristor elements on a single chip to convert 10 binary input data into 1,024 analog signals. This results in a structure that reduces the process of converting sensor data into digital form and transmitting it.

The research results showed that the technology demonstrated improved processing performance compared to existing artificial neural network (MLP) methods.

Data transfer volume decreased by about 10 times, and energy consumption was reduced by about 20 times. In addition, processing speed improved by 100 times.

It demonstrated accuracy levels of approximately 95% in noisy pattern recognition, 88% in 2D shape classification, and 99% in motion trajectory tracking. It also maintained an accuracy of over 95% in 3D object recognition and multi-wavelength detection.

Intelligent sensing systems consist of sensors, data processing units, and control units, and delays and power consumption occurring during data transmission have been identified as major challenges.

This technology operates using a 'proximity sensor computing' method that directly processes data at the sensor level. This is a structure that performs some operations on-site without transmitting data to a central processing unit.

The research team stated that the technology can be applied to fields requiring real-time data processing, such as autonomous driving, security, robotics, and smart homes.

Professor Seo Hyung-tak explained, “It is a new approach to compressing and processing data at the sensor stage.”

This research, supported by the Next Generation Intelligent Semiconductor Technology Development Project and the Mid-Career Basic Research Support Project organized by the Ministry of Science and ICT and the National Research Foundation of Korea, was published in the May issue of the international academic journal 'Advanced Functional Materials'.