How cities can get ahead with AI-enabled traffic management

Beyond accuracy, computer vision allows comprehensive 24/7 monitoring of video feeds, unlike human operators who can only monitor a few screens at a time. The result is quicker detection and response. These capabilities enable the AGIL Urban Traffic Management System to reduce incident detection time from minutes to seconds.

Taking incident detection beyond cameras

There are generally two approaches to incident detection in traffic systems. One is the approach already discussed, using cameras to visually monitor roads for incidents like accidents, stalled vehicles, or illegal stops. This works well at fixed locations like junctions or highways where cameras can be installed. However, it’s not practical to cover an entire road network, especially stretches that span hundreds of kilometres.

That’s where traffic data becomes very useful. Cities can collect real-time information from GPS-enabled probe vehicles like taxis or delivery fleets which provide a picture of traffic flow. Traditionally, this data has been analysed using rule-based methods – for example, comparing upstream and downstream traffic speeds or volumes, and looking for sudden drops or irregular patterns that might suggest an incident. These methods can be helpful, but they’re limited in fast-changing or unpredictable traffic scenarios because they’re based on fixed rules pre-defined by a traffic expert.

Cities can collect real-time information from GPS-enabled probe vehicles like taxis or delivery fleets which provide a picture of traffic flow

Artificial neural networks (ANNs) offer a more adaptive and intelligent solution. Inspired by how the human brain processes information, ANNs are made up of layers of interconnected nodes, called neurons, that can learn to recognise complex patterns in large datasets. In traffic management, platforms like our AGIL Urban Traffic Management System use ANNs to analyse real-time and historical traffic flow data, such as vehicle speed variations and traffic volume changes over time and across different locations. By training the network on known patterns of normal and abnormal behaviour, the system learns to recognise conditions that likely indicate an incident, even without any visual confirmation.

Unlike rule-based systems, neural networks don’t rely on fixed assumptions. They learn from historical and real-time data, which allows them to detect subtle changes and anomalies that traditional systems might miss. This makes them particularly powerful for wide-area incident detection – covering large zones beyond camera visibility – where quick detection and response can make a big difference in reducing congestion and improving safety.

From detection to decision

When a city builds up this level of intelligence in its transport systems using AI, it can unlock more intelligent coordination for critical services like emergency services and response planning. For example, AI can help emergency services find the fastest possible route to an incident. By using real-time traffic data and predictive models, AI can recommend the best path to avoid congestion or roadblocks. This routing can be dynamically updated as conditions change on the ground.

Beyond vehicle movement, artificial intelligence in transport networks can also support better incident response planning. Traditionally, traffic operators relied on standard operating procedures to decide the steps to take when there’s an incident – like calling an ambulance or notifying the traffic police. But traffic conditions are fluid, and effective response often require decisions that go beyond predefined protocols. For instance, transport authorities may need to adjust downstream traffic signals to divert vehicles away from the scene, or activate digital road signs to warn drivers in advance.

Beyond vehicle movement, artificial intelligence in transport networks can also support better incident response planning

Across a dense urban traffic network where there are hundreds of variable message signboards across expressways, notifying approaching vehicles isn’t enough – cities should also inform drivers on surrounding highways who are heading towards the affected area. Manually coordinating this messaging in real-time would be impossible.

That’s where the AI-powered AGIL Urban Traffic Management System comes in. The system helps operators automatically generate an appropriate response plan: which signs to activate, what messages to display, and where to display them based on each driver’s proximity to the incident. 

For example, a driver just 500 metres away needs specific lane information to avoid the hazard, while someone 10 kilometres away might simply be advised to take the next exit and avoid the route entirely. The faster cities can coordinate and act, the better they can manage traffic, prevent secondary accidents, and improve overall road safety.

Technology convergence for smarter mobility

Technologies like AI, IoT, as well as multimodal transport data are improving the efficiency of urban traffic management. Traditionally, detecting an incident relied on infrastructure like inductive loops embedded in the road. Today, with the integration of IoT and AI, cities have access to a wide array of real-time data – from connected vehicles, mobile devices, roadside cameras, and even crowdsourced information. When combined with AI, this wealth of data can be processed more intelligently and enable faster, more accurate decision-making.

In this context, AI is the brain of the ecosystem – it analyses the data, identifies patterns, and enables predictive decision-making. IoT functions as the nervous system, providing real-time sensing and feedback. Meanwhile, multimodal transport data provides a comprehensive view of how different parts of the mobility network – drivers, commuters, public transport – are interacting with one another. Together, these technologies enable more responsive, adaptive, and resilient traffic management systems.

At ST Engineering Urban Solutions, we work with cities to understand their pain points and leverage our domain expertise, experience and technology to address these effectively. We believe the future of urban traffic management is intelligent, integrated, and human-centric. That’s why we continue to innovate and iterate in partnership with cities, ensuring our solutions keep pace with both technological advancements and the real-world challenges cities face today.

ST Engineering is a global technology, defence and engineering group with a diverse portfolio of businesses across the aerospace, smart city, defence and public security domains. Harnessing technology and deep engineering expertise, its Smart Mobility business is advancing urban mobility, enabling cities to modernise their rail and land transport infrastructure for seamless, safe and reliable commuter journeys. To date, it has delivered more than 400 Smart Mobility projects in 90 cities across Asia, the Middle East and the United States.