The International Light Rail Magazine
+44 1733 367610

Tram automation: not if, but when?

With almost entirely segregated running and a high level of technological innovation, Dubai's tramway, opened in 2014, also features metro-like operational characteristics. Image credit: Hywell Waters © Panos-Rea/TAUT

Last April, in a front page splash, the Edinburgh Evening News informed its readers that the Scottish capital’s trams were set to go driverless as part of a plan to save money for future investment.

Not everyone noticed that the publication date of the article was the first of April and many responded to the hoax with predictable outrage – mainly over the presumed safety implications of automated tram operations. But, looking back, what seems surprising isn’t the hoax itself but the fact that driverless trams should strike anyone as outlandish enough to be an April-fool-worthy idea in the first place. Surely, given the current state of available technology and the economic benefits of dispensing with drivers, it is not just feasible to run a driverless tram network but inevitable that someone, somewhere will be doing it? So, where are they?

Look away from tramways for a moment and driverless technology seems to be popping up everywhere. Metro and shuttle systems (especially in airport transit lines) equipped for fully unattended train operation (UTO) are by now familiar to travellers all over the world. These trains do not need to be staffed at all – although some operators will choose to keep someone onboard – and they have been successfully tried and tested from Shanghai to Barcelona, from Paris to Dubai.

These services are controlled by technologies that have evolved over decades, since the London Underground’s Victoria line – the world’s first automatic passenger carrying train – and Paris Metro line 11 respectively opened in 1967 and 1969. Both still required a driver to look for obstacles on the track, respond to emergencies and operate the doors, but the automated train protection systems were already sophisticated enough to prevent trains passing signals at red into occupied ‘blocks’ of track and to safely control acceleration and braking.

Evolving the technology

Since the Victoria line and Paris line 11 blazed the trail, these technologies have progressed beyond all recognition. Whereas the first wave of automation worked on a simple static ‘block’ system, ensuring that blocks of track segregated by signals are only ever occupied by a single train or that an empty block should always be kept between any two trains on the same line (the belt and braces approach taken in New York and elsewhere), modern technologies permit a constant updating of information between track and train so that speeds can be adjusted according to real time conditions. The big gain from this, apart from safety, is that the safe distance between two trains on any given section of line can be kept at a minimum. This can translate to an increase in the number of trains running through a line by up to 20%.

Those sort of operational efficiency gains are as attractive to tram operators as they are to their colleagues on the world’s metros, so it isn’t surprising that some modern tram networks already use such driver assistance and protection technologies. But, unlike on a metro, what we don’t see are trams that take the final step to fully driverless operation. Why? The reason is deceptively simple.

Trams’ forced trade-off

A tram network is very different from a metro in one fundamental aspect: whereas a metro is a closed system, entirely separate from the rest of the city’s infrastructure in terms of its operations, the tram is not.

Trams are heterogeneous networks. They are interlaced with all other modes of transport that operate at surface level, from cars and buses to cyclists and pedestrians. Of course, some separation can be planned in, with trams passing under road or rail here and there and a degree of separation for stretches of the line – a tram is no longer thought of as a bus by other means – but an effective tramway must nonetheless be deeply integrated with other modes of transport, almost by definition. In fact, the multimodal aspect of a tramline, offering very quick and easy transfer from foot or cycle to tram to train, is often a major selling point for city planners. In these circumstances the challenges of providing unattended, driverless operation have to be looked at in a different light. A tram has to safely negotiate its journey with a constant stream of human and mechanical traffic, much of it crossing the line in more or less unpredictable ways, while still achieving optimal journey times.

It is not that we lack the technology to develop safe driverless tram services in complex city environments; as mentioned before, automatic train operation systems are already used in modern tram networks, ensuring signals are observed and speeds are appropriate for high risk parts of the route. We already have the tools, the sensors and beacons that would allow a driverless tram to detect unexpected obstacles and to respond appropriately. But for a tram there is a forced trade-off between safety and availability that a metro system does not have to make.

It is very easy, for example, to imagine a situation such as a pedestrian unexpectedly, but intentionally, crossing a tramline ahead against the signal where a driverless tram would have to respond with an emergency stop or service brake, but where a driver would be able to respond more appropriately to the situation, judging the danger with a sophistication that an algorithm cannot. It is easy to imagine because what would be a very unusual incident on the metro happens every day on tramways all over the globe. A safety protocol that was appropriate to every possible incident likely to be encountered by a tram would mean a disrupted and slowed service that would be inconvenient to passengers and very likely disruptive to other road users in turn. That’s if the company’s lawyers could ever be persuaded to sign it off in the first place.

Automation, but still with drivers

Perhaps these objections are overstated. After all, fully automated cars are constantly in the news and are being piloted in tests all around the world. There is a huge amount of investment and optimism (not to say hype) surrounding them and some glowing reports from early trials. Cars must be even more deeply integrated with urban traffic than trams; if they can operate safely on the roads, why can’t we do the same on rails? Well we can, in one sense, of course, from a technological point of view, but the private car or taxi does not have the same kind of service commitments that a mass transit service like a tram must operate under. Slow speeds, frequent stops and ultra-conservative risk assessment might all be practical for a downtown car journey through heavy traffic, but would render the tram service near useless. It is precisely because they want to avoid the inconveniences of car travel that our passengers are – at least some of the time – taking the tram in the first place.

Considerations such as these become very pressing when you are thinking about creating a tram system from scratch in a place that has never seen one before.
SYSTRA project manage the development of the Dubai tram, the first phase of which opened in 2014 and which the second and third phases have recently been awarded; if there is a city anywhere in the world likely to take the lead in a cutting-edge technology such as driverless trams, it is that one.

There had never been an urban tram system in the Middle East, so there were few preconceptions about what it could or couldn’t do, and Dubai takes great pride in demanding the best in terms of quality and performance. Local conditions, including extreme temperatures and sandstorms, meant that enclosed, air-conditioned stops with automated platform doors were essential if users were to be lured aboard. This, in turn, implied a high level of system automation so that the trams could quickly and reliably be brought to a stop in the correct position to align with the platform doors. In addition, there was the problem of creating from scratch a population of qualified drivers.

All the conditions in Dubai seem ideal for a driverless tram if ever there were to be one, and yet the tramway still has a driver in the cab. The trams and the lines bristle with innovation and cutting-edge tech, but the driver is still a human being because the fundamental problem persists that a fully integrated tram – and multimodal integration was an essential part of the Dubai project – still cannot operate at the levels of efficiency a modern city requires without the added component of human judgement.

Reassuring, perhaps, in an age where human redundancy seems to draw ever closer. Or perhaps not. Dubai still needs a human behind the handle, yes, but this tram system comes very close to the line that separates driver assistance from driver obsolescence.

CBTC technology is the communications-based train control system that exploits technologies that can be radioed to the train instead of using harder to maintain track circuits. The use of CBTC means that the tram’s automation can be more responsive to local conditions than ever before, with each tram being treated as a moving obstacle relative to all others, keeping headways down and throughput efficient. Information on each tram’s position is constantly updated in real time and the state of the tramway as a whole is constantly visible to operators for monitoring and analysis that is used to maintain high levels of efficiency.

The speed and responsiveness of CBTC has many current advantages, but it also points tantalisingly towards the possibility of a driverless future which – when it comes – will surely depend on the sort of rapid digital communications that have made CBTC solutions so effective in any mass transit developed over the last two decades. There is a ratchet effect when it comes to the application of technology that improves tram safety and speed control such as CBTC – once it has been employed it is hard for future projects to take a step back.

So never say never. The attractions to operators both in terms of cost and efficiency of an effective driverless system are huge, so the incentive to make the next step is there. But the complexity of the problem should not be underestimated either, and the temptation of false equivalence with other transport technologies, such as the driverless car should be resisted.

The allure of glamorous technological solutions is easy to understand, but maybe there is some satisfaction in the knowledge that so far the complexity of our city environments is beyond the understanding of our machines as well.

Certainly we will continue to advise upon all such developments that come along,
and if and when driverless trams become practical you can expect SYSTRA to be delivering them.