AI warnings cut illegal parking by 99%, study finds

- Korea Institute of Civil Engineering and Building Technology develops edge AI-based 'MOVI' system

- Displays license plates and nearby parking availability in real time

- Estimated to save about 20 million won per site annually

The illegal parking AI management system, MOVI, installed along a chronic illegal parking stretch in Byeollae-dong, Namyangju, Gyeonggi Province. [Korea Institute of Civil Engineering and Building Technology]

South Korean researchers have developed an AI-based system that tackles chronic illegal parking not through fines or enforcement, but by alerting drivers in real time and guiding them to nearby parking spaces. In field tests, the technology reduced illegal parking by up to 99%.

The Korea Institute of Civil Engineering and Building Technology said Tuesday it had developed the illegal parking AI management system — dubbed MOVI — which uses edge AI-powered deep learning video analysis to detect illegal parking instantly and encourage drivers to move their vehicles voluntarily.

A research team led by Lee Yu-hwa, a senior researcher at the institute's road and transportation research division, installed MOVI along a chronic illegal parking stretch in Byeollae-dong, Namyangju, Gyeonggi Province, and analyzed its real-world impact on traffic conditions.

The number of illegal parking incidents lasting five minutes or longer during peak commuting hours fell by up to 99% as of last August compared with September last year, before the system was introduced. Compared with January this year, when the system was in its early stages of operation, the figure dropped by about 85%.

The key distinction is guidance over enforcement. Unlike the conventional approach of issuing fines, MOVI assesses the situation on the spot and nudges drivers to move their vehicles on their own.

Edge AI cameras installed at the site analyze in real time whether a vehicle is parked illegally, with a detection accuracy of about 98%. When an illegally parked vehicle is identified, a variable message sign displays the vehicle's license plate number alongside real-time availability information for nearby public parking lots. LED floor lights installed along the curb also activate to prompt drivers to move.

MOVI deployed in the field, with a variable message sign displaying illegal parking alerts. [Korea Institute of Civil Engineering and Building Technology]

Data collected by the system is transmitted to a control platform, which authorities can use to analyze illegal parking patterns by time of day and zone and craft more effective traffic management policy.

The reduction in illegal parking has also improved road visibility, benefiting both vehicle traffic and pedestrian and driver safety. By directing drivers to nearby public parking lots, the system is also expected to boost utilization of existing parking facilities and ease the burden on enforcement personnel and administrative operations.

The research team estimated that MOVI would generate social cost savings of about 20 million won ($14,700) per site per year.

Building on the Namyangju pilot, the institute is expanding field testing and commercialization to other local governments this year, including Gimhae in South Gyeongsang Province, and plans to develop customized operating models tailored to the road and urban environments of each region.

"The core goal is to hit two birds with one stone — cutting parking management costs for local governments while improving road safety," institute President Park Seon-gyu said. "We will advance this into an AI city management technology that proactively addresses real traffic hazards, including securing golden-time access for emergency vehicles."

nbgkoo@heraldcorp.com