Artificial intelligence (AI) now scans all national highways across the country to identify high-risk accident zones and proposes tailored improvement measures. Starting July 7, the accident prediction service, previously applied to 5,900 km, will expand to cover the entire 14,220 km network, shifting road safety management from post-accident response to proactive prevention.
According to the Korea Transportation Safety Authority (KTSA) on June 6, the agency will begin a pilot service of 'K-Safer,' a traffic accident risk prediction and prevention solution, on all national highways under the jurisdiction of five regional Korea Expressway Corporation offices and connected roads. The coverage spans 14,220 km of national highways—approximately 2.4 times the previous 5,900 km.
In January 2024, the KTSA completed the K-Safer system installation on 5,900 km of national highways, 188.2 km of the Seohaean Expressway (Dongseocheon JC to Geumcheon IC), and approximately 878 km of major roads in Sejong City. From July 7, the service will extend to all national highway sections.
Traditional road safety management focused on post-accident responses. K-Safer, however, leverages big data and AI to predict accident risks and prioritize preemptive measures by eliminating hazards.
The system analyzes vehicle operation records, accident histories, road/facility data, traffic volume, speed, and weather information. The KTSA has accumulated around 1.3 billion records, including driving patterns and risky behaviors. It also incorporates data on icy road sections, variable message signs (VMS), speed cameras, and other roadside infrastructure.
AI first predicts accident likelihood by time slot, then identifies potential accident types and causes. It quantifies risks per vehicle and road length, categorizing sections into four levels: severe, caution, attention, and safe. Beyond identifying high-risk zones, it recommends improvements tailored to accident types and facility conditions.
A representative case is the Jeonmi Intersection in Wanju County, Jeollabuk-do. Located at the junction of National Route 21 and Science Road, this four-way at-grade intersection adjoins the Wanju General Industrial Complex. While the main route of National Route 21 passes under the intersection via an underpass, the ground-level access roads are curved, and guidance facilities are inadequate.
From 2020 to 2023, 10 accidents occurred here, injuring 18 people. Side collisions and rear-end crashes accounted for 80% of incidents (four each). Five additional accidents occurred in 2024. K-Safer classified the human risk factor at this intersection as 'severe,' the highest level, with speeding, signal violations, and lane deviations as primary causes.
To mitigate these risks, AI generated 35 improvement proposals, including overlapping measures. Short-term solutions included installing enforcement cameras, colored pavement markings, and vehicle barriers. Long-term plans suggested geometric redesigns to address the irregular intersection layout. The system not only quantifies risks but also provides actionable solutions.
Monthly traffic safety reports are generated for each high-risk section. The KTSA and road authorities conduct joint inspections and use these reports to develop improvement strategies.
Risk information is also shared with drivers via VMS and civilian navigation apps like TMAP, encouraging safe driving near hazardous zones. Road managers receive data to justify facility upgrades, while drivers get real-time risk alerts.
The system has proven effective: accident counts in K-Safer pilot zones decreased by 19.1% compared to pre-implementation levels. A KTSA official stated, “We aim to establish an AI-based traffic safety management system that proactively eliminates risks before accidents occur. Through monthly safety reports and joint inspections with road authorities, we will support the implementation of targeted improvement measures.”
