Daegu Transit Corporation has begun preparing a field demonstration of an artificial intelligence system intended to identify and help prevent crowd-related accidents in railway stations. The corporation signed an agreement on November 4 with participating organizations, including Anna, to test the technology at Banwoldang Station, a transfer station served by Lines 1 and 2. The project is a development assignment selected under the 2026 Public Innovation Demand-Based Innovative Product Technology Development Program led by the Public Procurement Service and the Korea Institute of Procurement Research.
The demonstration is scheduled to run from July 2026 through June 2028, with a total budget of 1.3 billion won. As the organization providing the test environment, the corporation will support equipment installation, operate the field demonstration and provide feedback from station operators. The available evidence describes a technology validation project, not a completed safety system or a confirmed reduction in accidents.
Twelve demonstration CCTV cameras will be installed in major congestion areas, including the concourse, transfer passages and platforms. AI video analysis will estimate the number of people in each zone, measure crowd density and examine movement flows. Those results will be connected to a three-dimensional model reflecting the station’s actual layout and to pedestrian simulations. The combined system is intended to identify congested sections and bottlenecks created by concentrated passenger movement, then assess whether crowd risks can be forecast before conditions become more dangerous.
The planned system will operate on a separate closed network that is physically isolated from the corporation’s existing CCTV and business networks. External internet connections and remote access will be blocked as part of the information-security arrangements. Through June 2028, the corporation plans to evaluate the performance and practical applicability of AI video analysis and digital-twin technology in an operating subway environment. It will then consider whether the results could support a pilot purchase and possible expansion to other stations.
