Efficient commute with Smart-Q
SMRT-Q is a conceptual prototype that leverages real-time crowd sensing and advisory systems to reduce MRT platform congestion, improve commuter experience, and support Singapore’s Smart Nation mobility vision.
Context
As part of Singapore’s Smart Nation push, smart mobility initiatives aim to improve daily commutes. With a growing population and curbs on car ownership, reliance on public transport—especially MRT—continues to rise. This project focuses on enhancing the MRT experience through technology-driven solutions for crowd management.
Problem
Despite measures like shorter train intervals and staff deployment, MRT platforms and trains remain overcrowded during peak hours. Commuters face uncertainty in choosing where to queue, often clustering near escalators and doors, worsening congestion. SMRT also incurs high manpower and operational costs in managing human traffic flow.
Goals
Therefore, there is a need to create an Information System (IS) that can strategically complement the current initiatives by leveraging information technology (IT) to offer a better travelling experience for the commuters.
We will do this by providing better advice on emptier carriages and opening more doors for future SMART initiatives in transport and mobility. These efforts would provide opportunities in managing the crowd congestion in trains and on platforms, and point out the weaknesses of the co-existing efforts that can be improved, based on organisational learning on commuters’ activities and behaviours.
To design an information system that:
Provides SMRT with commuter data for operational and strategic planning.
Eases platform congestion by distributing crowds more evenly.
Advises commuters on emptier carriages in real-time.
Case Studies: Assessed existing “traffic light” systems for crowd indication, noting limitations in influencing commuter decision-making.
Surveys: 100+ respondents shared dissatisfaction with crowding.
List of questions included in the survey:
Which station do you find is the most congested?
How often do you take the train during peak hours?
How satisfied are you with the train schedules?
How crowded are the platforms during your journey?
Rate your experience during congestion period.
How often do you take the train per week?
How long do you usually take to board the train during peak hours?
Which sections of the station do you usually move to board onto the trains?
Do you think it will be helpful to implement this system?
Would you take advisory from the light indicators on the MRT platform if this prototype was implemented?
Field Studies: Observations across Jurong East, Bishan, City Hall, and other interchanges revealed commuters bunching at escalators and ignoring announcements to move in.
Benchmarking: Compared computer vision solutions (OpenCV vs Camlytics) for real-time passenger counting; OpenCV chosen for flexibility, cost-effectiveness, and accuracy.
We defined success as reducing commuter waiting time by at least 1 train interval (≈2–4 minutes) during peak hours.
Alignment was achieved across three levels:
Commuters: Faster boarding, reduced stress.
SMRT operations: Lower manpower costs, more efficient crowd distribution.
Management: Richer data for planning new lines, train frequency, and station upgrades.
We designed SMRT-Q, a prototype advisory system with four key components:
Real-time counting (via CCTV + OpenCV facial/head detection).
Real-time feedback (LED indicators: green = empty, amber = moderate, red = full).
Remote feedback (data relayed to the next station and stored centrally).
External database (for organisational learning, analytics, and planning).
Camera setup
SMRT-Q works in 5 simple steps
The SMRT Closed-Circuit Television (CCTV) will take real-time images at regular intervals to provide updated information on carriage occupancy levels.
The image is processed by the application on the server on board the train. The application counts by detecting the number of people.
The server then connects to the Wi-Fi and sends the data to the server at the next station and to an external central database for other SMART systems.
The station processes the information and sets the light colours of the indicators according to the pre-set threshold.
The data is then stored in the external central database for organisational learning and higher-level decision-making.
Design considerations included:
Minimal cognitive load → simple traffic light system that commuters can instantly understand.
Scalable integration → plug into existing SMRT infrastructure.
Future-proofing → potential use for security, station crowd heatmaps, and predictive planning.
We proposed a pilot launch at Bishan MRT during peak hours, chosen for its heavy but not extreme congestion. Key measures of success:
Reduction in average waiting times.
Distribution of commuters across middle vs end carriages.
Uptake rate: % of commuters following indicator advisories.
Expected quantitative improvements:
2–4 minutes saved per commuter during peak hours.
>60% compliance from survey respondents willing to follow indicators.
Reduced staff intervention, lowering operational costs.
Conceptual prototype validated through surveys, field observations, and feasibility tests.
Strong commuter acceptance: 80% rated SMRT-Q as helpful, 60% would follow the indicators.
Clear organisational benefit: data from SMRT-Q could inform train frequency planning, station upgrades, and long-term transport policy.
Simple, human-centred solutions (traffic light indicators) can meaningfully influence commuter behaviour.
Data-driven insights (from carriage occupancy trends) empower SMRT to plan operations and long-term transport strategies more effectively.
Pilot studies are essential to validate feasibility and refine adoption strategies before scaling.
Designing at a systems level means addressing the needs of commuters, operators, and policymakers simultaneously.