Deepfine 与首尔交通公社合作开发 AI 有轨电车安全平台

가

The body content of the article will be changed to this letter size.

스크롤 이동 상태바

Deepfine, a spatial AI-based industrial platform company, is partnering with Seoul Metro to develop an AI-powered tram safety platform that identifies and analyzes potential operational hazards.

Deepfine said on Oct. 20 it is developing AI-based tram track hazard detection technology under the 2026 AX Innovative Enterprise Creative Technology Development Program, supported by the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP).

The project aims to build a tram safety platform that uses AI to analyze the movements of pedestrians, bicycles, vehicles and other objects near tram tracks in real time, enabling operators to predict and respond to potentially dangerous situations.

As the lead research organization, Deepfine will develop core technologies including multisensor hazard detection and analysis, 3D spatial modeling, a digital twin-based control system and smart glasses designed to assist tram drivers.

Seoul Metro will establish hazard classifications and operational standards based on its expertise in tram operations and railway safety. It will also participate in field demonstrations on the Wirye Line.

Research and development began in April and will continue through December 2027, including subsequent research phases subject to interim evaluations.

During the first year, Deepfine will develop AI technology that combines data from multiple sensors, including CCTV cameras and light detection and ranging (LiDAR) systems, to identify hazards near tram tracks.

The system will analyze the trajectories, speeds and directions of individual objects to detect potentially dangerous movements, including approaching or entering tracks and sudden movements toward them.

Deepfine said the technology will assess risk by combining information such as an object's location, distance from the tracks, movement speed and time to collision (TTC).

Based on the analysis, the system will classify risk into three levels: safe, caution and danger. Even in crowded environments with multiple pedestrians and vehicles, it is designed to continuously track individual movements and assess the risk associated with each object.

The resulting hazard information will be transmitted to the control system and tram drivers to support on-site responses.

Deepfine will also develop technology that delivers hazard warnings, speed reduction recommendations and other operational information to drivers through smart glasses.

The aim is to help drivers identify hazards that may be difficult to detect visually, enabling faster decisions and responses using sensor data and AI.

In subsequent research phases, Deepfine plans to enhance its risk analysis technology using time-series and spatial data and establish an integrated 3D digital twin-based control system.

The system will recreate actual tram routes in a three-dimensional environment, visualizing the locations and movements of pedestrians, bicycles and other objects, as well as hazardous areas, in real time.

It will also analyze warning signs such as increasing congestion, objects approaching tracks and abnormal movement patterns over time.

Deepfine said the technology will allow control center operators to assess conditions and prioritize risks, supporting preventive action.

Field demonstrations will be conducted in stages alongside development.

Deepfine and Seoul Metro plan to validate the first-year multisensor hazard detection system along the pedestrian-heavy transit mall section between Wirye Lake Park and Wirye History Park.

Subsequent demonstrations will evaluate the enhanced risk analysis technology and 3D digital twin control system.

Deepfine has set performance targets of at least 95% accuracy in detecting hazardous objects and situations, a response time of no more than 0.5 seconds between hazard occurrence and the display of warnings on control screens and smart glasses, and a detection rate of at least 90% for hazards outside the driver's line of sight.

The company plans to commercialize the tram safety platform using technology and operational data obtained from the Wirye Line demonstrations.

It also intends to expand applications beyond new tram routes to urban rail systems, including subways and light rail, as well as smart cities and public safety.

Deepfine plans to build a common technology base for spatial recognition, risk assessment and on-site response support, while exploring potential applications in industries such as logistics and manufacturing.

Hyun-Seon, Park hspark@thelec.kr

My comment