ETRI WEBZINE

ETRI researchers have announced the development of traffic signal optimization technology using Artificial Intelligence (AI), which has reduced average travel time for vehicles by more than 15%. This advancement paves the way for smoother urban traffic flow.

Electronics and Telecommunications Research Institute (ETRI) has announced the development of the ‘City Traffic Brain(UNIQ^*)’, a signal optimization technology that applies artificial intelligence reinforcement learning. This technology effectively adjusts traffic signals quickly in response to changing traffic volumes.

Optimizing traffic signals at small-scale intersections often leads to a balloon effect, where improvements in one area cause deteriorations in another. However, ETRI explained that the cloud distributed processing technology developed this time enables easy application of this technology even in over 200 large-scale intersections.

ETRI has constructed road networks and a signal system for over 1,500 intersections by receiving data from Daejeon, Sejong, and T-map (T-map). Furthermore, to enhance signal optimization performance, video information collected from approximately 800 cameras was analyzed using deep learning technology. Traffic volume was estimated with less than a 10% error rate^** using data generation technology based on their proprietary traffic simulation technology.
* UNIQ: Urban Network Intelligence for solving traffic Queues
** The meaning of a 10% error rate: It refers to the difference between the actual measured traffic volume and the simulated traffic volume using estimated vehicle demand, calculated as the Mean Absolute Percentage Error (MAPE).

Previous traffic signal optimization technologies did not consider constraints such as pedestrian crossing times, rendering them impractical for real-world application. However, ETRI’s City Traffic Brain technology accommodates all signal constraints, including ensuring pedestrian crossing times and maintaining the sequence and cycle of signals for the safety of drivers practicing predictive driving, making it immediately applicable to actual roads. ETRI has proven its applicability in reality by implementing this technology in traffic lights at ten intersections in Yuseong District, Daejeon Metropolitan City.

The City Traffic Brain technology is provided through a cloud platform, allowing for phased expansion to other local governments. Additionally, it is expected to improve the quality of life for individuals by alleviating traffic congestion, resolving transportation inconveniences, and enhancing the air quality environment.

Furthermore, the ability to rapidly respond to changes in road conditions, such as lane changes, new road constructions, or the erection of large-scale facilities, will also aid in the formulation of scientific traffic policies.

ETRI cites the core technologies of this development as including ▲traffic signal optimization technology using AI, ▲cloud-based large-scale traffic simulation technology, and ▲edge-cloud collaborative traffic situation analysis technology.

This technology involves equipping traffic lights with simple signal control devices, where data collected through traffic analysis collectors and video information are processed by edge servers and integrated with a smart city control center to control signal optimization via an online system.

Chung Moonyoung, a principal researcher at ETRI’s Smart Data Research Section, stated, “Based on the developed technology, we plan to conduct further research on environmental impact assessments of traffic volume changes due to external environmental factors, such as the relocation of administrative offices or the opening of new buildings.”

Kim Tae-soo, the Director of Traffic Policy Department in Daejeon Metropolitan City, also stated, “Changing the traffic signal system is the most effective way to improve urban traffic congestion. We will make an effort to reflect the outcomes of this joint research in policies aimed at improving traffic flow.”

Kim Chang-hyun, an official from the New Transportation System Team in Sejong City, explained, “Sejong City is being developed with the goal of becoming a public transportation-centric city. As the traffic situation changes frequently with the occupancy of different living zones, utilizing the developed technology will greatly assist in swiftly optimizing large-scale networks and relieving traffic congestion.”

The research team plans to transfer technologies such as ▲traffic simulation software (SW), ▲traffic signal optimization modules, and ▲traffic demand data generation tools to companies involved in traffic-related enterprises or smart intersection projects, with the aim of commercialization within the next year. Through this research, they have achieved significant outcomes, including the filing of 13 domestic and international patents and the publication of 28 papers domestically and internationally.

ETRI has stated that for the rapid commercialization and application of this technology, integration is necessary, including the databasing of maps, signals, traffic volumes, and traffic-related maps, as well as the utilization of data owned by local governments and correction technologies. Further research will be conducted to expedite system construction.

This work was supported by the Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korean government (MSIT) (No. 2020-0-00073, Development of Cloud-Edge based City-Traffic Brain Technology). Participants in this four-year project included KAIST, InnoGrid Co., Ltd., LEXGEN Co., Ltd., NAVER System Co., Ltd., Modutech Co., Ltd., Daejeon Metropolitan City Hall, and Sejong Special Autonomous City Hall.

Chung Moonyoung, Principal Researcher
Smart Data Research Section
(+82-42-860-6599 mchung@etri.re.kr)