Learned Programme

For many of us  the experience of crossing at busy junctions makes us feel second class to those passing in front of us in motor vehicles. Although the Highway Code now places pedestrians at the top of the road user hierarchy, many junctions are yet to be upgraded to provide adequate pedestrian priority and still largely focus on the movement of vehicles. But how can pedestrians be given greater priority without making congestion, and air quality, significantly worse? How can the needs of bus users be balanced with those of pedestrians? In this presentation we will draw upon examples from a range of authorities to illustrate how computer vision detection is enabling authorities to make more nuanced decisions about when to prioritise pedestrians whilst also minimising the impact on the network.

Traffic congestion and queues on urban road networks cause high delays for car drivers, buses, and also some pedestrians. Traffic congestion and queues also generate air pollution which affects all road travellers and also some non-travelling residents. Traffic control strategies seek to reduce at least some of these large negative impacts. However, selecting the right control strategies is not a straightforward task, since road networks are complex by nature, and their state depends on capacity-demand interactions, travel behaviour (including routeing decisions), congestion dynamics and other dynamic and stochastic factors. In this paper we consider upstream-gating merge-control. This strategy systematically protects specific sub-networks from long queues and congestion without reducing overall network throughput. This paper contributes to the state of the art in network-wide traffic control, taking explicit careful account of users’ response to control policies (including routeing changes) in a rather simple way. The paper demonstrates in a network example both

• how the traffic equilibrium is affected by upstream-gating merge-control and also
• how the dynamical evolution of traffic is affected by upstream-gating merge-control.

The paper shows how the upstream-gating merge-control can be easily tuned to reduce public transport delays.

Background; Traffic routeing and traffic control. The paper very briefly reviews the literature on traffic routeing: traffic routeing models estimate and predict the traffic flows and travel times on a road network, given an input demand and road capacities (including green-times) offered. The paper also very briefly reviews the literature on traffic control. The literature on both routeing and control is vast so these reviews will be very brief. Smith et al. (2012, 2019, 2023); especially 2023; may be regarded together as the start point of the present paper.

“Upstream-gating” control strategies. This paper focusses on “upstream-gating merge-control” strategies which hold traffic back at traffic signals just ahead of a merge to prevent the formation of a queue at a downstream bottleneck; and we also consider different ways of further, additionally, controlling the two inflows to the merge. We show, by considering a simple example network, that if the upstream-gating merge-control uses only flows to control the two approaches to the merge then delays may get arbitrarily large as travellers swap to cheaper routes. For example, this happens with the “zipper” rule (which equalises the two inflows at the merge). On the other hand, we show that adding upstream merge-control which equalises the two delays felt at the merge does not have this effect and so maximises the network capacity, notwithstanding the upstream gating. This suggests that, in general, delays should probably be used to control merges and downstream queues, rather than only flows, if overall network capacity is to be maximised. This observation may help the design of good control strategies, using both flows and delays, for upstream-gating designed to remove or reduce queues at specific downstream locations. We include a dynamic analysis allowing for the dynamic growth of queues in the example network. The analysis here makes reasonable allowance for the spatial extent of queues but does not consider within-cycle or cycle-to-cycle queueing dynamics. The paper will also show how the “equal delay” rule at the entrances to the merge may be varied to yield much smaller delays on one approach to the merge than the other approach; this may be employed to reduce public transport delays. The upstream-gating merge-control strategies we consider in this paper automatically take reasonable account of consequent traffic re-routeing; they maximise network capacity.

Prizes: Work on traffic control and routeing, reported in the papers listed below, was awarded the 2007 Robert Herman Lifetime Achievement Award by the Informs Transportation Science & Logistics Society; and in 2021 this work was awarded the 16th Kometani-Sasaki prize.

Some background papers; the italicised are referred to directly in this abstract.

Smith, M.J., 1980. A local traffic control strategy which automatically maximises the overall travel capacity of an urban road network. Traffic Engineering and Control 21, 298–302.

Smith, M. J. 2012. Traffic control and route choice; modelling and optimisation. Presented at the JCT Symposium, University of Warwick (September 21, 2012). Available from the JCT website.

In Rotherham, we were successful in securing nearly £600k in funding through the TSOG awarded by DfT. Using our existing Imtrac data, it was quickly realised that all of the asset along the A631 in Rotherham was life expired. The basis of our bid was to upgrade the obsolescent equipment along this route and try to future proof the route and make it ready to test equipment crucial to bringing Connected and Autonomous Vehicles (CAVS) to the region. Early contractor involvement meant the project was started in April 2024 and completed in November 2024!! Not only did we upgrade what we set out to do, we even found some hidden bounces! What were these? You’ll have to wait for the presentation! 

A look into how the largest roundabout in Plymouth overcame detection issues with smartmicro above ground sensors.

The Swansea Bay & West Wales Metro is a regional initiative aimed at providing reliable bus services across the South West Wales Authorities. By implementing traffic signal intelligence, the project seeks to enhance the speed and reliability of bus journeys, thereby reducing travel times across the region.

In line with the Department for Transport’s LTN 1/24: Bus User Priority guidance document, this paper outlines the approach taken by Swansea Council, in collaboration with AECOM, during the design and implementation stages of the Swansea Metro project. It details the research carried out for various technologies and innovative design processes enabling a comparison of Local & AVL bus priority systems.

In 2021, York became the first UK city to deploy real-time traffic modelling within its traffic management operations. The main benefit of this deployment was to fuse together live data from the city's assets such as: timings from signals connected to UTC; and traffic flow detection from loops and cameras, with city-wide floating vehicle data and a transport model to provide an accurate prediction of current and near-future traffic conditions. This information allows the city to be proactive in its approach to traffic management.

In 2024, the city has implemented a real-world pilot to automatically simulate alternative signal plans within the real-time model, and to send recommendations to the network management officers to implement via the UTC. The combination of automated simulations - triggered by user-defined thresholds and results from the near-future traffic predictions - and implementation of the plans recommended using a customisable scoring system, resulted in traffic delay reductions of up to 8% in the peaks. 

In our paper and presentation, we will explain the architecture of the system, present the results of the switch-on / switch-off test, and summarise the benefits and lessons learned from the pilot. We will also consider the options and potential benefits for city-wide rollout of such an approach.

Designers of traffic signal junctions spend plenty of time examining the theoretical capacity, but as the importance of active modes of travel increase do we fully understand how many people a new facility might serve?  This paper examines the theoretical capacity of a crossing phase to see how many pedestrians could cross in a single cycle, because there might be times when that is important.

The FLEX detector is a self-contained, stand-alone detector designed for traffic data collection, that requires no power and no side road cabinets. Based on advanced AI algorithms and models, provides exceptional performance and 99% accuracy. Data is sent to cloud based servers through a built-in LTE modem, and can be retrieved through a user friendly web interface or API. We'll present a case study from the Swedish Road Administration where the FLEX is used for capillary data acquisition in secondary and rural roads. 

A practitioner’s guide for incorporating passive safety into traffic signal designs, including dispelling myths and misconceptions, and the problems still to be faced.

This paper outlines a groundbreaking project led by Kirklees Council to control the flow of pedestrians and vehicles through a busy, multi-lane crossing in Huddersfield. Working with Starling Technologies, the Council are using predictive AI techniques to overcome pedestrian crowding and jaywalking at a central, high footfall location. This paper will present the baseline data and the understanding of crowd dynamics at the roadside. It will go on to detail the interventions put in place to adjust traffic and pedestrian flow during peak times, highlighting the impact on safety.