Kakao Navi AI Learns Inconvenient Routes for Better Navigation, Acceptance Up 6.65%p

Kakao Mobility has opened new horizons in navigation by developing AI route guidance technology that learns drivers' actual reactions. This innovative technology, which transcends the limitations of existing navigation systems, has been internationally recognized for its excellence with a paper published in 'TRC (Transportation Research Part C: Emerging Technologies)', an SCI-level academic journal authoritative in the transportation field. Until now, navigation systems have relied solely on physical information such as vehicle speed, road width, and number of lanes to provide route guidance. However, actual drivers often deviated from the guided routes for various reasons, such as illegal parking zones, inconvenient entry/exit ramps, or distrust of unfamiliar roads. Kakao Mobility focused on these "temporary deviations" by drivers. They introduced a method that analyzes drivers' "route compliance rate" – how well drivers followed the navigation's guided route – using AI to re-evaluate the "traffic value" of roads and reflect it in route search. This was made possible by independently developing an AI algorithm based on 'Multi-armed Bandit (MaB)', a methodology within reinforcement learning. Thanks to this technology, navigation systems can autonomously learn the inconvenient factors that influence drivers' route choices. It allows for continuous improvement in navigation quality without the need for separate infrastructure, and it significantly enhanced the accuracy and reliability of route guidance by precisely calculating the traffic value of millions of road segments nationwide and reflecting real-time traffic information. In practice, this technology detects inconvenient sections that traditional navigation systems struggled to reflect, such as areas prone to frequent route deviations due to queues like the segment from the southern end of Dongjak Bridge to the Olympic Expressway entrance, complex alleyways, congested areas near transfer centers, and mountainous roads with significant elevation changes, and reflects them in real-time guidance. This technology, applied to Kakao Navi since November last year, has demonstrated remarkable performance improvement. According to the paper, data analysis from the initial week of technology application showed that drivers' route compliance rate for 'fastest routes' surged by a significant 6.65 percentage points, from 64.22% to 70.87%. Kakao Mobility stated that this research, based on driver behavior data, provides improved routes by reducing the 'discrepancy' between navigation information and actual driving environments, achieving significant results across various route quality indicators such as reduced driving time and increased driving convenience. Now, Kakao Navi will offer a more pleasant and reliable mobility experience with smart guidance that even understands the driver's mind.