A new concept for traffic signal control

When police control traffic, they look at the traffic – not at the clock. Self-controlled traffic signals work the same way, just much faster and more precisely. Here, the signals themselves determine when each traffic stream gets to go and for how long. Sensors detect how many vehicles are approaching from each direction and the optimization algorithm continuously recalculates signal timing to minimize stop time for all vehicles.

1 Problem
2 Concept
3 Methodology
4 Case Study
5 Literature

1 Problem

Controlling traffic at intersections is a difficult and important task. It makes a big difference whether one can move freely through the city or wastes time stopped in traffic. This applies to motorists, public transport users, cyclists and pedestrians alike. Treating all intersection approaches equally would be efficient if traffic flowed uniformly from all directions. But demand varies, buses can arrive late, traffic patterns change. No transport planner can precisely predict traffic demand. The best way to smooth traffic flows is to react flexibly to actual conditions. And this is exactly how the self-controlled traffic signal system works.

2 Concept

At a self-controlled traffic signal transport planners do not specify the order and duration of green signal phases, instead they simply define the main traffic properties (e.g., which sensors measure which traffic streams, minimum green times, safety times, etc.). These properties define a framework for understanding the traffic’s dynamic behaviour. The self-controlled traffic signal uses this framework to, for example, assign one traffic stream more green time than another, or combine traffic streams flexibly to provide pedestrians priority in one instance and trams priority in another. In short, to do whatever the traffic situation requires.

The self-controlled traffic signal system works by preparing a robust traffic forecast and using a powerful optimization method to set green times. The system regulates traffic flows to minimize waiting time over the next two minutes based on current traffic conditions. Since these conditions can change quickly, for example when a pedestrian pushes a request button, the optimization and decision-making process is repeated every second.

Single node:
The self-controlled traffic signal (top) controls the two traffic flows by minimizing waiting times and stops. The fixed time control (bottom left) gives green for exactly 30 seconds to each direction. The time gap control (bottom right) switches to red whenever there is a large gap.

The self-controlled traffic signal generates green waves automatically, because stopping a large vehicle column would be assessed poorly in the optimization process. Pedestrians are given a green signal with every major gap in vehicle traffic. If pedestrians have been waiting a long time, smaller gaps are accepted and, if no gaps are available, a few cars are delayed for a short time. The self-controlled traffic signal gives buses and trams priority by assigning them a higher weight (user determined) in the optimization. If the conflicting traffic volumes are higher the public transport vehicle will need to wait, but otherwise it will get green on arrival, then the traffic signal will return to assigning green time to the other traffic flows according to their urgency. In short, the traffic signal behaves as if it were controlled by the traffic rather than controlling the traffic itself.

Dynamic traffic stream management:
The self-controlled traffic signal can dynamically manage the large traffic volumes on the curved main road as well as empty the secondary roads quickly. The fixed-time control system (bottom left) creates a green wave. The time gap control system (bottom right) forces large columns of traffic to stop frequently.

The self-controlled traffic signal system’s is more reliable because the signal times are not fixed, but are adapted to suit the needs of current traffic volumes. Furthermore, due to the system’s shorter phase times, significantly more (and consequently smaller) traffic columns move through the network. Pedestrians and cyclists benefit from these shorter columns by receiving more frequent green phases, and public transport vehicles stop less frequently.

Harmonious public transport prioritization:
The self-controlled traffic signal system establishes a harmonious cooperation between cars and public transport (top) by giving public transport vehicles a weighting of 15 cars. In this case when there is a large volume of opposing traffic then the bus needs to wait for a short time. If one were to treat the bus just like a car (bottom left), then it would be stuck with the cars. With a weight of 100 (bottom right), traffic on the main road would stop.

The self-controlled traffic signal system successfully prioritizes public transport almost without impacting traffic flow. In other words, giving priority to public transport does not necessarily lead to a deterioration in service quality to other transport modes.

3 Methodology

The self-controlled traffic signal system is based on a simple methodological structure: a powerful optimization algorithm determines the best possible signalling sequences based on an estimate of vehicle arrivals created using traffic volume data for each network intersection and a set of transparent model building blocks. The algorithm seeks to minimize the weighted sum of wait times and stops over the next two minutes. Transport planners no longer need to decide exactly what the traffic signal should do in all situations. Instead, by selecting the target weights, they specify what are the most important traffic flows.

  • Simplified planning: Transport planners only need to describe the intersection qualities, not the duration and sequence of green phases. This simplifies planning, because these parameters do not need to be adjusted if traffic volumes change over time or due to diversions. Traffic is not managed according to pre-configured signal plans, but rather depending on the actual situation and needs. The self-controlled traffic signal system takes traffic as it comes.
  • Harmonizing special requirements: The self-controlled traffic signal system allows planners to specify what should be done in special situations, for example when a pedestrian requests a green phase to cross the street. This situation should not always be treated the same way: for example, if a tram is in the intersection the pedestrian may receive a weight of 5 to ensure that she can cross the street to catch the tram. With the self-controlled traffic signal system, such requests can be easily formulated. The system can balance the requirements of traffic streams against each other and resolve them in a compatible manner.
  • Demand-based control: The self-controlled traffic signal system reacts automatically to different traffic demands. When less traffic is coming from one direction, it has more green periods available for the other directions. Under full load the system attempts to empty the queues one after the other by providing long green periods. When there is less traffic, the optimization process has more freedom to, for example, prolong the green for stragglers, or provide more frequent green phases for pedestrians. In periods of very low demand, such as late at night, practically no one has to stop.
  • Robustness: The self-controlled traffic signal system uses virtual vehicles to prevent errors. The system monitors detectors continuously. As soon as the evaluation electronics or the logical plausibility check detects an error, the detector measurements are ignored and replaced by modelled traffic volumes. For example, if ten vehicles per minute are expected on average, ten virtual vehicles are simulated instead of using data from the failed detector. Thus, even in the event of a fault, each traffic stream gets sufficient green time.

4 Case Study

A prototype of the self-controlled traffic signal system was tested at two adjacent intersections on the Königsbrücker Landstraße in the north of Dresden [Google Maps, Bing]. These intersections were chosen because they were manageable in terms of size and complexity, and because they include several complicated traffic features.

Location: The test area extends over two intersections on the North-South axis of Dresden. The figure illustrates the vehicle detector sensors (blue) with their distance to the stop line, the streetcar tracks (yellow) and the crossings for pedestrians and cyclists (red).

Originally, the two intersections were operated with a traffic-dependent VS-PLUS control system. This system is based on day-time-dependent frame plans with 80 or 100 seconds of cycle time. The signalling is coordinated with four additional intersections via common green bands in both directions. In contrast to VS-PLUS, the self-controlled traffic signal system does not have a supporting program. The traffic situation at the intersection alone determines which traffic streams will get green when and for how long.

Dresden Case Study:
The self-controlled traffic signal system provides free flow for the trams and buses. The requirements of pedestrians, cyclists and motor vehicles are met flexibly and in accordance with requirements.

The results of the case study confirmed the potential for harmonious public transport prioritization and dynamic traffic flow management, as shown in the previous simulation studies [pdf]. Instead of coordinating a set of intersections, the self-controlled traffic signal system optimizes traffic flow based on the actual traffic situation at each individual intersection. In a quantitative comparison with a traffic-dependent VS-PLUS control, buses and trams experience almost no delays, pedestrians and cyclists wait a third less time for the next green, and the motor vehicles experience lower waiting times and the same high throughput rates although the traffic volumes were 10% higher.

Summary Results:
The self-controlled traffic signal system significantly improved traffic quality for all modes of transport.

The self-controlled traffic signal system can reliably solve the complex problems and requirements of real traffic networks. The system uses robust estimation models to compensate for erroneous measurements with simulated traffic volumes and allows users to input specific data including start-up delays, distance behaviour or travel speeds. In the case of signal group control, specific requirements have also been implemented, such as the RiLSA-compliant release of tolerable flows, the safe guidance of blind people via central islands, the signalling of railway crossings, and providing dynamic time islands for passenger changes at stops.

The Dresden case study results show that self-controlled traffic signal systems are a sustainable and scalable solution for efficient and low-maintenance operation of traffic signal systems under the real requirements of today’s cities. It establishes a new building block for innovative, efficient and future-oriented urban transport management.

The case study was financially supported by the European Regional Development Fund (ERDF) and the German Research Foundation (DFG Project Tr 1102 / 1-1). The project continued to be supported by Dresdner Verkehrsbetriebe AG the Dresden road and civil engineering department, Siemens AG, the Professur für Verkehrsleitsysteme und -prozessautomatisierung as well as the Professur für Bahnverkehr, öffentlicher Stadt- und Regionalverkehr at TU Dresden.

The idea of "Self-Controlling Traffic Lights" was honoured with the Golden Idea Award of IDEE-SUISSE® - Swiss Society for Ideas and Innovation Management in 2012 [video].

5 Literature

  • T. Karrer, M. Bartsch, A. Genser, S. Lämmer, C. Heimgartner (2021) Praxistest in Luzern zeigt: Selbst-Steuerung verbessert die Verkehrsqualität deutlich. Straße und Verkehr 5/2021, 10-23. [www]
  • S. Lämmer (2016) Die Selbst-Steuerung im Praxistest. Straßenverkehrstechnik 3/2016, 143-152. [pdf]
  • E. Lefeber, S. Laemmer, J.E. Rooda (2011) Optimal control of a deterministic multiclass queuing system by serving several queues simultaneously. Systems and Control Letters 60, 524–529 [www]
  • S. Lämmer (2010) Prinzipien der Selbst-Organisation vernetzter Verkehrsströme. In: Bahn, Bus, Pkw - Optimierung in der Verkehrsplanung, Hrsg: M. Krüsemann, M. Schachtebeck, A. Schöbel, Seiten 19 - 42, Shaker. [www]
  • S. Lämmer and D. Helbing (2010) Self-Stabilizing Decentralized Signal Control of Realistic, Saturated Network Traffic. Santa Fe Working Paper Nr. 10-09-019. [www] [pdf]
  • S. Lämmer, J. Krimmling, A. Hoppe (2009) Selbst-Steuerung von Lichtsignalanlagen - Regelungstechnischer Ansatz und Simulation. Straßenverkehrstechnik 11, S. 714-721. [pdf]
  • S. Lämmer and D. Helbing (2008) Self-Control of Traffic Lights and Vehicle Flows in Urban Road Networks. Journal of Statistical Mechanics: Theory and Experiment, P04019. [www] [pdf]
  • S. Lämmer, R. Donner, and D. Helbing (2007) Anticipative control of switched queueing systems. The European Physical Journal B 63(3) 341-347. [www] [pdf] [video lecture]
  • S. Lämmer (2007) Reglerentwurf zur dezentralen Online-Steuerung von Lichtsignalanlagen in Straßennetzwerken (Dissertation, TU Dresden). [www] [pdf]