Journey time prediction for bus priority at traffic signals

Many City Authorities in Europe are now seeking to tackle increasing congestion and pollution through demand management and improved public transport. An important contribution to this strategy is the provision of public transport priority, both through physical segregation and through other priority techniques. In the U.K., effort in recent years has focused particularly on the provision of bus priority at traffic signals, using a variety of techniques including gating, pre-signals, queue relocation and selective vehicle (bus) detection (SVD). With SVD, priority is provided to individual detected buses by extending or recalling the green signal, subject to the constraints of the priority strategy.Most of the development and implementation of bus priority using SVD in the U.K. has occurred in London. The first major implementation occurred in the SELKENT bus district, where 900 buses were fitted with transponders to activate priority at 56 isolated junctions. Evaluation (University of Southampton, 1987) indicated an economic payback period of about one year and has led to system expansion so that 300 isolated junctions now offer SVD bus priority. More recently, active bus priority has been developed for the SCOOT UTC system within the European Commission funded DRIVEII project PROMPT (Hounsell and Landles, 1995). This has been implemented in London at 19 junctions in the Camden Town and Edgware Road SCOOT areas, with field trials indicating an economic payback period of around 16 months. Following these successful trials (Hounsell et al, 1996), bus priority has been included in the latest version of SCOOT (V3.1) released recently by TRL (Bretherton and Bowen, 1996). Development of bus priority in fixed time UTC areas is now underway in London, following similar principles.Bus detection in London has been based on bus-mounted transponders interacting with bus detectors which are sited typically 70m-100m upstream of the stopline. This location has been shown to allow reasonable accuracy in journey time prediction from detection to clearing the stopline whilst also allowing reasonable response time for the signals, assuming the priority request can be actioned immediately (i.e. at the local controller, rather that at the UTC centre). A longer detection distance is desirable where the link is "dean" (giving consistent predictable journey times) or where central control is operated (where a 4-5 second communications delay occurs). Detection is always downstream of bus stops, due to the unpredictable waiting time involved, so that a detection distance less than 70m is sometimes necessary with a consequent reduction in bus delay savings.Bus priority of this form has also been introduced in other U.K. cities, such as Southampton, where "footprint" detection has been used (no bns-mounted equipment is then required). Bus priority has also been trialled at some junctions in Southampton using radio-based Automatic Vehicle Location (AVL) for locating buses approaching traffic signals, with this information being passed directly to the UTC system for priority activation. (Buses using a trial corridor in Southampton have been fitted with AVL as part of the STOPWATCH system for real-time passenger information at bus stops.) In this case, buses are located every 20-30 seconds according to the radio polling cycle, rather than at a fixed point, and the potential exists for implementing selective priority according to bus status (adherence to schedule, occupancy, etc.).For given bus, traffic and site characteristics, two key factors determining the effectiveness of bus priority at traffic signals have been shown to be (i) detection location (through inductive loops or AVL) and (ii) journey time predictive accuracy. These factors have provided the focus of a project undertaken by the University of Southampton for the Engineering and Physical Sciences Research Council (EPSRC) under grant GR/J92439. This paper summarises the activities and achievements of the research.