Researchers at Korea University, Dongguk University, the Korea Institute of Science and Technology (KIST), the Korea Photonics Technology Institute, and the Pohang Accelerator Laboratory have developed a new near-infrared vision system that, they say, can see through even thick fog — and could be used in future autonomous vehicles for better safety.
“This ultra-low-noise organic light sensor enables precise obstacle detection even in dense fog, making it ideal for vision-assisted systems in autonomous driving, medical imaging, and security,” says co-corresponding author Min-Chul Park of the team’s work. “Its compatibility with flexible substrates and low power consumption allows deployment across various platforms, from vehicle exteriors and road infrastructure to drones and smart traffic systems-overcoming the limitations of conventional sensors.”
Researchers have tested (a) a new camera system capable of picking out objects (b) in heavy fog (c). (📷: Oh et al)
Park and colleagues had previously developed an artificial intelligence (AI) algorithm that could process video feeds to help reduce the impact of fog and other vision-obscuring phenomena — but its latest approach is implemented entirely in hardware, taking the form of a self-assembled electronic blocking layer dubbed “3PAFCN,” which when integrated into the organic photodetector, reduces noise and increases sensitivity.
In lab simulations, designed to test its performance in conditions approximating real fog, the sensor proved able to capture enough data to reconstruct the shape of target objects at light levels far below where rival visible-light sensors failed. This could be improved still further, the researchers suggest, by combining it with the team’s earlier AI algorithm.
The team’s work has been published in the journal Advanced Materials under closed-access terms.