SK Planet Wins Prime Minister's Award for AI Road Surface Detection Technology 'ARHIS' Based on Driving Noise

SK Planet's in-house developed ARHIS (Audio & AI based Road Hazard Information System) won the Prime Minister's Award at the 18th Meteorological Industry Awards. This technology uses AI to analyze the noise generated when a vehicle drives on the road, detecting hazardous road surface conditions such as black ice in real time.

How It Works and Key Technical Features

ARHIS operates based on driving noise and image data collected via built-in microphones and cameras. The collected data goes through a preprocessing process and is converted into feature values in the form of MFCC (Mel-Frequency Cepstral Coefficients), with wind noise and environmental noise filtered out to improve quality.

The AI model has a two-stage inference structure. In the first stage, it classifies the basic road surface condition (Dry, Wet, Icy, etc.), and in the second stage, it applies an anomaly detection method to determine whether freezing has occurred (Wet/Icy) in wet road conditions. Through this, the system ultimately achieves road surface condition determination at an accuracy of 99%.

Advantages Over Existing Technology

Unlike existing physical sensor-based methods, since it uses sound, installation location is flexible and maintenance is easy. In particular, it is unaffected by low-visibility environments such as nighttime, fog, and fine dust, allowing stable recognition of road conditions. In addition, its wide detection range means it works effectively even in special environments such as bridges and tunnels.

Business Status and Future Plans

ARHIS is currently installed at around 400 locations in Korea, providing data to local governments and transportation agencies. Going forward, SK Planet plans to expand the scope of road safety monitoring by adding auto-labeling technology, in which the AI model directly labels data, and a fog detection feature.