Why predictive control is urban mobility's future
‘Adaptive traffic management represents the most significant advancement in how we handle urban mobility challenges’, begins Matthias Künzelmann, Strategic Program Manager at SWARCO.
‘In the 60s and 70s, the solution was simple: build more roads, add more lanes. But we now know this actually attracts more traffic. It's a vicious cycle.’
Künzelmann explains why adaptive traffic management has become essential. ‘You can't keep building forever. We need to be smarter with what we have. Moreover, the world has become more complex. In the past, it was only about vehicle flow. Now we're dealing with buses, cyclists, pedestrians, and increasingly diverse stakeholders – all with their own, often conflicting interests. For example, giving priority to buses for better service usually means longer wait times for individual transport. Or if you create longer crossing times for pedestrians, it comes at the expense of traffic flow for public transport. And if you try to reduce emissions by limiting stops for trucks, that can increase waiting times for cyclists.’
The technological context driving adaptive traffic management also changes rapidly. For example, the start-stop systems and electric vehicles change our old assumptions about emissions. At the same time, new possibilities are emerging through connected mobility, floating car data, and more powerful detection systems with intelligent cameras. ‘And as if that's not enough’, continues Künzelmann, ‘we have a shortage of traffic engineers because many young people expect artificial intelligence (AI) to solve all this automatically. Mobility experts are still needed to train AI, define constraints and ensure safe operation with all that new complexity.’
The evolution of traffic control
To understand why adaptive systems are so powerful, Künzelmann urges us to take a look at how traffic control has evolved: ‘First-generation systems worked with fixed time plans – preset cycles for different traffic conditions during a day that remained the same regardless of actual traffic conditions. These traditional systems are still seen worldwide, but they can't respond to the dynamic nature of traffic.’
The third generation brought real adaptability. ‘These systems adjust traffic lights directly, based on real-time measurements. They look at what just happened and modify the next cycle – reactive, but still not forward-looking.’
Predictive control as a breakthrough
The real breakthrough came with the fourth generation that uses predictive control. Instead of just reacting to what has already happened, these systems use sophisticated traffic models to predict what will happen in the coming seconds or minutes.
‘Imagine five cars approaching from one direction and two from another’, he illustrates. ‘A predictive system can simulate different scenarios and determine that it's smarter to prioritize the larger group because this results in fewer overall stops. The system decides based on what's going to happen, not what just happened.’
A multimodal approach
Modern systems are no longer focusing solely on cars. ‘In the early days, we only looked at cars, but today's cities are complex. Predictive systems can recognize all traffic participants and include them in their decisions. The system can see if a large group of cyclists is approaching, or if a bus is coming that should get priority.’
This multimodal approach aligns perfectly with many cities' vision on mobility. SWARCO can give structural priority to cyclists or public transportation with their systems, making these sustainable modes of transport more attractive. These algorithms adapt to different goals.
How it works in practice
Modern adaptive systems combine various data sources for a complete traffic picture. ‘We don't just use traditional detection methods like loops and cameras, but also new sources like floating car data and information from connected vehicles’, explains Künzelmann.
‘With what we call Model Predictive Control, or MPC, the system continuously simulates different traffic scenarios for an upcoming time interval’, he continues. ‘It evaluates various possible control patterns based on specific goals – such as fewer stops, lower emissions, or priority for public transport – and then chooses the best solution.’
Challenges and opportunities
‘The initial investment in predictive systems is slightly higher than for traditional systems’, acknowledges Künzelmann. For municipalities with tight budgets, this can be an obstacle, although the hardware often costs the same. ‘It is more the missing willingness or fear for innovation. But the long-term benefits – less congestion, lower emissions, fewer accidents – deliver a solid return on investment.’
He also points to the unexpected advantage that predictive systems actually make configuration easier: ‘Traffic engineers can focus on setting priorities – such as pedestrians over cars, or public transportation over private vehicles – and the system takes care of the optimal implementation.’
The future
In the future, model-based approaches can be further enhanced using digital twins, foresees Künzelmann. Instead of analysing isolated intersections or corridors, modern technologies enable predictive, near real-time simulation of entire cities. These digital twins integrate not only adaptive traffic signal control, but also variable message signs and dynamic speed regulations to simulate and optimize traffic flow across the entire network—empowering data-driven, strategic mobility decisions on a city-wide scale. This broader perspective opens up new possibilities for more environmentally aware traffic management, as the factors contributing to air quality issues often lie beyond the immediate location where they are detected—and may be influenced elsewhere within the network.
