Effective Detours and Curtailments

Planned Versus Spontaneous Curtailments, and the Role of Machine Learning

TfL’s iBus system has a comprehensive, proven curtailments module specifically designed for London, which manages situations where a route is required to end early and the vehicle instructed to turn around, cutting off the last portion of the outward and inward trips.

In most cities, curtailments can occur at any stop. In London, however, due to the size and complexity of the city and its large vehicles, TfL uses the concept of ‘curtailment points’ where buses can be safely and efficiently turned.

To meet TfL’s requirements, Trapeze has implemented the concept of curtailments within iBus: for each curtailment triggered, iBus ensures consistency of curtailment information across buses, service controllers and passenger information.

This system enables services to continue in the event of disruption, maintaining accurate prediction information through the remainder of the route and, very importantly, ensuring the operator is paid in line with the service they were asked to deliver, rather than the originally scheduled service.

This requires curtailment information to be passed from the real-time tracking system to the London Reporting Database (LRD) in a reliable and auditable manner.

This model was specifically designed for London, where there is a shared responsibility for curtailments. TfL takes decisions to curtail routes through CentreComm, which are then implemented by the bus operator. The entire process is underpinned by a sophisticated radio messaging system that enables dispatchers to effectively target affected vehicles.

Machine Learning and Spontaneous Curtailments

In other, less complex cities, curtailments are usually managed more spontaneously than they are in London. This is at least partly because few authorities have the requirement or resources to undertake such a complex and comprehensive planned curtailment network as has been delivered here.

However, while there is little doubt that London’s tailored solution is both hugely effective and highly dependable, are there elements from elsewhere that we could consider including, to the benefit of passengers and Transport for London?

One of the potential concerns for TfL could be that the present approach is dependent upon the skill and knowledge of a small number of CentreComm dispatching experts. We believe that machine learning could be used to augment the present solution, empowering these dispatchers to make consistently effective decisions, while simultaneously reducing reliance on individuals.

In practice this would involve using algorithms to dynamically present CentreComm with detour options in real time, chosen through a combination of present situation and historic data. Dispatchers could retain control, but their decisions would be based on information and forward simulation, substantially reducing the element of human guesswork.

While this concept is new in the bus sector, something similar is part of the product suite of one of Trapeze’s sister companies, Signature Rail, based in York. Bus networks may be more complex and more subject to external influence than rail ones, yet we believe the same principles could be applied here, empowering dispatchers to make consistently efficient, informed decisions.

In addition, machine learning offers the potential for curtailments to be applied with far greater precision. For example, a machine-assisted approach could make it possible to choose a range of diversion routes from the same point on a trip, in a way that ensures effective headway management across all vehicles. This would contrast with today’s approach, where typically the same diversion is utilised by all vehicles on a route.

Finally, one of the possible lessons to be learnt from other cities, which may be applied to London, is that curtailment instructions could be delivered electronically, thereby reducing voice traffic.

Let’s Discuss!

While London’s present curtailments solution is a highly customised and dependable solution, we believe there may be potential to increase the efficiency of services by building elements of machine learning-driven spontaneous curtailments upon the present foundations.

As ever, we at Trapeze aim to support Transport for London through the continued evolution of the iBus solution. By considering the latest technology developments, and experiences of other major cities around the world, together we can ensure that London remains a global flagship bus network and ITS solution.

Interested in discussing the potential for evolving detours and curtailments within iBus? Let’s talk!