SMRT Turns MRT Gantry Data Into an AI Crowd Radar for Proactive Train Deployment

Singapore's rail network moves millions of commuters a day, yet crowd management has largely been a rear-view mirror exercise — operators study the data and respond after congestion forms or a disruption hits. SMRT wants to flip that sequence with a commuter pattern analysis system that pairs artificial intelligence with MRT gantry data to predict where crowds will build before they actually appear.

The system, announced on Wednesday (Aug 5), learns commuter travel routes from faregate data and tracks how those patterns shift over time, mapping hotspots in real time across the network — most often at interchange stations. SMRT says the insights will let it manage crowds proactively and optimise train deployment, ramping up frequency where demand is about to spike. When disruptions do occur, AI-driven route recommendations would steer commuters to alternatives quickly and help the operator anticipate where demand will surge next. The idea isn't new: SMRT studied Guangzhou Metro's operations through the SMRT-Guangzhou Metro Rail Innovation Laboratory, a joint facility opened at Bishan Depot in June.

The commuter analysis work sits alongside a broader digital twin push. SMRT is trialling a Smart Station at Marina Bay after operating hours, building a real-time virtual model of the station — fed by sensors, CCTV, lifts, escalators and air-conditioning systems — so a single dashboard can flag problems early. Digital twin models of the North-South and East-West line track and tunnel infrastructure are also in development for sharper defect detection. The thread running through all of it is the same one visible across Singapore's transport tech: shifting from reacting to failures to anticipating them, from predictive rail maintenance on the JARVIS platform to Changi's autonomous wheelchairs.

For commuters the payoff is fewer surprise delays and faster recovery when things go wrong — and the financial stakes are real. SMRT Trains revenue rose more than five per cent to $969 million in FY2025/26, with margins improving to 1.3 per cent, after the operator had to spend over $10 million on service recovery and repairs following the six-day East-West Line disruption in September 2024. That single incident is a reminder of what reactive operations cost; reinvesting the improved earnings into AI is SMRT's answer.

Why it matters for Singapore: As the rail network grows, engineering capability has to keep pace, and predictive AI is becoming the core of that capability. Feeding gantry data, cameras and sensors into decision-making systems will also test how Singapore balances operational efficiency with commuter privacy — a conversation the country is already having elsewhere in smart-city policy. But the direction is clear: the network of the future runs on prediction, not reaction, and SMRT is positioning itself at the front of that shift.