JULY 10 2018 was a landmark day in the history of railway operations. After a decade of development, and two years later than originally planned, Rio Tinto operated its first fully-automated driverless freight train on the 280km heavy-haul line between its mine at Tom Price and the port of Cape Lambert in Western Australia.

Three locomotives, all without drivers, hauled the train, which was carrying 28,000 tonnes of iron ore. Remote monitoring was conducted at Rio Tinto’s operations centre in Perth, more than 1500km away.

Operating automated and driverless trains is of course nothing new. The first automatic metro train entered service on London Underground’s Victoria Line in 1968. Drivers remain in place to open and close the doors, reflecting the different grades of automation (GoA) now in use.

The world’s first entirely automatic and unattended metro line (GoA 4), the Port Island Line in Kobe, Japan, opened in 1981. Since then driverless operation based on Communications-Based Train Control (CBTC), a key enabler of automation, has gradually become the standard for new metro projects and upgrades to existing lines, apart from in China where the merits of full automation have only recently come to the fore. According to the International Association of Public Transport’s (UITP) Automated Metro Observatory team, the length of fully automated metros has grown to more than 1000km in 42 cities in 19 countries around the world, reaching this milestone with the opening of Shanghai’s Pujiang Line in April. GoA 4 is now in place on 7% of all metros.

However, transferring this to the mainline is extremely complex. Unlike closed and point-to-point metro lines, main lines are defined by mixed traffic while issues with technological interoperability pose major challenges. There is also the question of working with different operating companies and infrastructure managers, which require multiple contracts. Critically, for GoA 3 and 4, many of the key requirements, such as new technical solutions for obstacle detection and platform protection as well as legal parameters have largely yet to be defined.

While a closed network, the scale of the Rio Tinto Autohaul project means it is a major step forward. The system installed by Ansaldo STS uses automatic train operation (ATO) over ETCS Level 2 at GoA 4, and ultimately will link 16 iron-ore mines with four ports. Rio Tinto’s success is already encouraging other mining operations to follow suit. Pilbara rival, BHP Billiton, is currently implementing a similar driverless project, this time under a contract with Bombardier. The foundations for the railway’s moving block system, which is not based on ETCS, is a 4G LTE communications network and commissioning is expected next year.

That Rio Tinto, and now BHP, are willing to pursue such complicated and expensive projects - Autohaul is estimated to cost $US 940m in total - is testament to the benefits on offer.

Automated operation promises increases in capacity by reducing headways and improvements in efficiency by enhancing timetable stability and punctuality. Unattended Train Operation (UTO) will also increase operational quality by eliminating the variations present in manual driving, reduce energy consumption by 20% as well as cut maintenance costs.

While Rio Tinto’s GoA 4 project is at the top end of current installations, there are signs that main line applications are set to become more commonplace with suppliers touting the benefits of “quick-win” ATO projects based on GoA 2 using existing signalling infrastructure.

Managing interfaces such as level crossings was one of the challenges overcome by the Rio Tinto AutoHaul project.

China is leading the way with the deployment of ATO on main line passenger networks. Among the early projects is an installation developed by Casco Signal, a joint venture of Alstom and China Railway Signal & Communication Corporation, for the Dongguan - Huizhou - Zhaoqing line.

China Railway (CR) looked to ATO to maximise onboard capacity through the installation of platform screen doors on the route in the Pearl River Delta, which experiences high traffic volumes. Operating over CTCS Level 2, the system provides GoA 2 capability, and according to Mr Thomas Bruel, Alstom’s signalling solutions director, the project is meeting CR’s requirements.

He adds that there is further appetite for similar ATO applications in China, including on its high-speed network. “In the short-term we see significant developments in China mostly based on CTCS levels 2 and 3,” Bruel says.

Russia is also making significant headway with various automation projects, including moving block operation at GoA 3 on the Moscow Circle Railway (MCR), which opened in 2016. The system was pioneered on the Sochi - Adler - Krasnaya Polyana line built for the 2016 Winter Olympics and seeks to increase capacity. Here a single operator controls up to 10 trains with real-time monitoring provided by a GPS-based positioning system embedded in the onboard safety system. Audio frequency track circuits support mixed traffic operation of passenger trains with freight services, offering two modes of operation: lineside signals for freight trains of an agreed weight and length and cab-based signalling for the suburban services which operate at headways of 2min 40s. In addition, digital radio channels are in use for data transmission of warnings to the ES2G Lastochka EMU fleet.

Elsewhere, Russian Railways (RZD) is also developing automated coupling using satellite-based technology while trials with automated and driverless shunting technology and traffic management systems are taking place at Chelyabinsk.

East Japan Railway (JR East) is reportedly exploring the use of ATO, with pilot projects proposed for the Yamanote Line in Tokyo and on the Tohoku Shinkansen. Here trains will initially run with a supervisor onboard with GoA 4 a long-term goal. However, the project faces significant hurdles, including installation of enhanced obstacle detection as well as grade separation and installation of platform screen doors at stations. A number of legal changes are also required.

With European-based suppliers at the forefront of some of the major projects that have been implemeted so far, Bruel and others believe that Europe will lead the technological development of future automated and driverless projects, which will be based on ETCS.

The commercial use of ATO over ETCS has been pioneered on London's Thameslink network.

The deployment of ATO over ETCS on the core section of London’s Thameslink network by Siemens, which has increased capacity to 24 trains per hour per direction, is a major breakthrough and is establishing a precedent for future projects. Similar applications and trials of ATO technology are underway in Germany, the Netherlands, Switzerland and Austria (see below).

French National Railways (SNCF) is also working on ATO development to boost the railway’s profitability. “Autonomous trains will make for smoother, more seamless rail operations, with trains synchronised and operating at the same speed,” says Mr Luc Laroche, head of SNCF’s Autonomous Train project. “This in turn means greater reliability and more on-time arrivals.”

SNCF is already conducting twice-monthly trials in the Paris region during the day with a test freight locomotive and a converted Corail test coach fitted with telecommunications tools, a computerised vision system and 3D recording equipment. Laroche says a key objective of the project is to provide engineers and experts with live operating experience of ATO-equipped trains.

SNCF is working with external partners on the tests, including leading signalling suppliers, new contributors from systems engineering, industry start-ups, and academic institutions through SNCF’s agreements with the SystemX and Railenium research institutes. In addition, Larouche says SNCF is cooperating with DB to forge a pan-European approach to autonomous trains, develop shared standards and design “the train of the future.”

Bruel says work on the 24-month SNCF project began in April 2017 and teams have already achieved a significant amount. He expects the first results from test benches and some of the line trials to be made public by the end of the year, followed by a full report in April 2019. SNCF is also set to hold a press conference on September 12 to outline further details of the work. “The goal of the programme is to identify what the key mission of the driver is with the system and how we can minimise this by using new technology,” Bruel says. “The idea is to create a collective and open innovation programme in order to collaborate better in such research initiatives by research institutes, SNCF and individual suppliers such as Alstom.”

Long-term vision

While early deployments have predominately focused on GoA 2, the long-term vision remains to develop GoA 3 and 4 main line applications. Existing ETCS infrastructure is suitable for supervised applications. However, more complex applications will require new technologies to support key functions such as obstacle detection, accurate train positioning, and enhanced voice communications.

A typical ATO system has three levels: a driver assistance system, which provides situational awareness and is able to detect what is happening on the train and the track in order to make the right operational decision; an auto pilot system, described as the brain of the operation, which provides decisions during the journey and facilitates the shift from acting upon instructions to developing its own rules; and a remote control system, which offers direct and continuous communication from the train, even in a remote location.

In a recent lecture to the Institution of Railway Signalling Engineers (IRSE), Dr Josef Doppelbauer, executive director of the European Union Agency for Railways (ERA), described how the rail industry is entering its fourth-generation signalling application, which will introduce new principles of operation.

For Doppelbauer this system is defined by trains collectively understanding the safe operating distance between different vehicles and to communicate wirelessly to a traffic management system (TMS) situated in the cloud. On the trackside the only element that will remain is the switch controller and the occasional level crossing. The switch setting can also be controlled by the train, enabling quicker release following the passage of the train. The onboard system is defined by a single interface with a bi-directional communications channel to the outside world while geographic interface determines the braking curve. Sensors installed on the train can monitor the condition of drivers, such as whether they have consumed alcohol.

To operate this system, Doppelbauer foresees a structure with three distinct layers, or loops of control. The innermost control loop is for each individual vehicle, and is equivalent to ATP installations today. This is followed by a collective layer, where vehicles communicate with each other, as demonstrated in work taking place in Shift2Rail on virtual coupling, which could lead to train platooning, with individual trains running much closer together. “This collective can also include non-rail vehicles [such as road vehicles to support Mobility as a Service], maintenance vehicles, and the passenger on the platform,” Doppelbauer says. The final layer is the overall network-wide TMS.

“As we will have a lot of intelligence, and a lot of intelligence on multiple trains, we will have a lot of redundancy in the system,” he says. “This redundancy will avoid single points of failure and will increase the reliability and the availability of the system.”

Sophisticated driver assistance systems, and sensor technologies as well as advanced data analytics and Artificial Intelligence are touted by suppliers as key technologies needed to deliver the fully-automated operations structure outlined by Doppelbauer. These technologies will support the high levels of infrastructure safety required. And with the road sector developing similar solutions as part of work on autonomous vehicles - SystemX is also actively researching automated driving applications - there is potential crossover. Indeed, Doppelbauer says the architecture for the fourth-generation network is identical to that under development for road applications. He believes this could enable rail to join the technology mainstream, improving competitiveness, and allowing it to remain the backbone of a multimodal transport system. It may also spark a shift in the supply market.

“The automotive sector is developing a lot of concepts in this area and from recent experience it is important to be inspired and able to take some of these onboard,” Doppelbauer told IRJ. “One major example of this is the radio technology used and the sector’s communications networks, which is most likely to be a 5G network. 5G will be fundamental to the future communications networks and future automation. But to achieve this, the railway sector will have to work with suppliers from other sectors, and these suppliers, and others, may be interested in coming into the rail sector directly in the future.”

Existing rail suppliers are certainly not unaware of this threat and recognise the need to embrace new solutions. Many are already diversifying their offers in order to deliver the applications railways are increasingly demanding.

Bruel says Alstom’s recent acquisitions of 21Net and Nomad Digital, companies which he describes as leaders in their respective fields, will strengthen the supplier’s offer in the onboard connectivity. Similarly, Thales’ purchase of Danish driving assistance provider Cubris in May, which is helping clients including Danish State Railways, South Western Railway in Britain, and Transdev in Sweden, to reduce energy consumption and train performance through its Greenspeed DAS system, is described by Mr Alberto Parrondo, Thales’ strategy director for ground transportation, as key to its future offer. He says the technology will be the backbone for autonomous trains.

Likewise, the $US 215m acquisition of Guavus, United States, a pioneer of real-time big data analytics, in 2017, supports the Thales’ digitalisation strategy and according to Parrondo will facilitate the use of Artificial Intelligence in future railway applications, including operations. He adds that the supplier is benefitting from technologies developed in other sectors of its business.

“From our aerospace and defence expertise, we are getting the sensors required for these projects,” Parrondo says. “Railways have not necessarily used things such as radar, lidar and satellite positioning before, but we have access to expertise from defence where there is plethora of knowledge about this. We are able to merge data from these sectors into a single relevant source. Our products are also cybersecure by design and we are investing more resources in this area.”

Alstom and Thales and other key suppliers are similarly immersing themselves in the latest research initiatives. Bombardier is working with Unisig and Shift2Rail, and touts the benefits of the programme to produce an interoperable approach to developing ATO solutions that are scalable to different applications.

Siemens is also engaged in Shift2Rail. “We believe that the key to success of the next generation of signalling lies in a joint approach to developing new ideas,” says Dr Rüdiger Brandt, director of sales, main line railway automation, at Siemens Mobility.

Shift2Rail itself is focusing its efforts through its Innovation Programme 2 (IP2), which is looking at advanced traffic management and control systems to develop and validate a standard ATO system up to GoA 3/4. Specifically, this includes technical demonstrators on a new communications system, ATO, moving block, train positioning, train integrity, a new laboratory test framework, standardised engineering and operational rules, virtual coupling, an optimised TMS, smart radio-connected wayside objects, and cyber security.

Ms Lea Paties, Shift2Rail’s programme manager for IP2, says automatic operation is a key component to achieve the programme’s primary goals of enhancing capacity and reliability. The initiative is divided into two workstreams and is making good progress with establishing standards for future ATO deployment. Paties reports that Shift2Rail achieved a major milestone in June with the first draft of defined specifications for ETCS and infrastructure to operate ATO at GoA 2 on main line applications.

“We have worked closely with the sector, but in particular ERA as the system authority, to get a formal renewal of the ERTMS system specification,” Paties says. “The goal is to get the specifications into law in four to five years and it has been critical to work with the agency to get this right from the beginning.”

The document remains a draft and not a formal update of the TSI, which is expected to take place in 2022. The next step is to develop a prototype GoA 2 system, with the goal of introducing a prototype suitable for a pilot test line by the end of 2018 or early 2019.

Alongside the GoA 2 project, work is taking place to develop the solutions and technologies required for effective GoA 3 and 4 main line applications. Lab tests are expected to take place in Britain in cooperation with Network Rail from 2019 to 2020, while line testing is expected to take place with DB in Germany in 2021-22. Ultimately, this will combine work in various workstreams with significant developments expected ahead of the programme’s conclusion in 2023-24.

A key focus for Shift2Rail is to retain ERTMS as the core technology, and where necessary to provide backwards compatibility to protect investments both in mainline and urban railways. However, beyond the developments that Shift2Rail and other initiatives are promising, increasingly the focus is likely to be on developing the next generation of ETCS. ETCS was not conceived with automation in mind. And while Level 2 is supporting the development of GoA 2 applications, it is more feasible both practically and economically for railways to achieve GoA 3 or 4 via ETCS Level 3, which will offer moving block operation.

Dopelbauer says an opportunity exists to accelerate the development of ETCS Level 3 through a hybrid system and he is encouraged by the work of Network Rail in this area.

“On the basis of these new technical capabilities, we need to review these development and deployment plans because in my view there are opportunities to accelerate towards Level 3,” Doppelbauer says. “If we introduce some of these generation-four features into ERTMS Level 3 we can come closer to the vision for the future. I believe then railways could migrate directly without investing in components that could become obsolete so we could move quicker in the future.”

Similarly, a new telecommunications platform will be required for such a system as 2G GSM-R simply does not offer the bandwidth required to support multiple and complex communications. BHP’s 4G application might then prove to be a landmark project for future applications. Rail will need to fight to secure the level of bandwidth it requires for these applications. Standardisation, while respecting market functionality, is also critical here and in other areas to avoid the interoperability problems of the past.

“If rail is to survive, it must do more to address the question of cost, which can be achieved through greater standardisation,” Doppelbauer says. “The rail industry needs to work closer together, which they are showing in Shift2Rail. We see the importance of continuing this level of collaboration, because without these structures it is unlikely to happen, and that is why we are currently looking to build political support for Shift2Rail 2.”

Significant progress has clearly been made with automated operations in recent years. Major railways now recognise that the benefits of ATO experienced in the metro sector are within reach of main lines. Light rail applications are also gaining traction (p76). However, with other sectors also exploring the benefits of automation, which have the potential to alleviate environmental concerns and congestion, their traditional drawbacks, rail cannot afford to stand still.
Standardisation and interoperability remain major challenges, but they have not gone unnoticed by the leading institutions working on these projects. The importance of initiatives such as Shift2Rail, SNCF’s work with System-X and Railenium, and DB and SNCF’s work to bring the key players together cannot be understated. The industry is positioning itself to overcome these difficulties and ultimately deliver a solution that is compatible with existing infrastructure but scalable to include future advances in technology.

There will undoubtedly be bumps in the road - will unions be ready to accept that the role of the driver will change significantly and potentially be eliminated in the longer term? But the stage is set for significant developments over the next few years as railways around the world advance various projects, and suppliers step up with the required solutions. The 2020s then may well become the decade in which driverless trains come to the fore, keeping rail relevant in an automated world.


ATO projects spring up across Europe

FOLLOWING hot on the heals of Thameslink, Paris RER Line A is set to benefit from an ATO system developed by Alstom under a €20m contract with Paris Transport Authority (RATP). The line carries 1.2 million passengers every day, or 70,000 per hour per direction, and the new system will control pairs of five-car M109 double-deck EMUs and older M12N trains on the central core. “The feedback from our customer and drivers has been very good,” Bruel says. “We are on target to complete the project by the end of 2018 and 70% of the fleet is already equipped.”

German Rail (DB) and the City of Hamburg announced plans in July to develop a fully-automated S-Bahn line in partnership with Siemens. Four trains will operate on ATO over ETCS Level 2 on the 23km eastern section of Line S21 between Berliner Tor and Aumühle, with drivers onboard for operation beyond the automated section. The goal is to introduce the line into service by October 2021 with a view to eventual full automation of the entire Hamburg S-Bahn network.

DB is also exploring main line applications. Former DB chairman Dr Rüdiger Grube estimated in 2016 that autonomous main line trains would be introduced by 2021-2023, and DB is pushing ahead with tests. It has allocated a 30km section of track on the Erzgebirgs Line in Saxony to trial DMUs with ATO, with the first tests taking place in mid-2017. The company’s ETCS rollout programme, adopted as part of the DB 2020 strategy, is also being conducted with automation in mind.

Swiss Federal Railways’ (SBB) has introduced the SmartRail 4.0 ATO programme, which is targeting the introduction of ATO GoA 2 by 2021-22 with subsequent upgrades to GoA 3/4 foreseen. The programme is taking place alongside the wider deployment of ETCS across the network, with complete onboard installations set to be completed by 2025, and SBB has already successfully tested a trackside module system (ATO-TS). Siemens and SBB were set to conduct their first joint pilot test of a GoA 2 semi-automated train management system at the end of August.

The introduction of ATO on the 100km Betuweroute freight line in the Netherlands from the port of Rotterdam to the German border near Emmerich as well as for shunting operations at Valburg yard, is another key European application.

Alstom is working with Dutch infrastructure manager Prorail and freight operator Rotterdam Rail Feeding (RRF) on the project to introduce a GoA 2 application where drivers supervise but do not drive or start the train. The application will utilise ETCS levels 1 and 2, which was installed on the line by Alstom around a decade ago, and is intended to increase capacity and reduce energy consumption. The contract for the project was awarded at the end of 2017 and Bruel says he expects the first results by the end of this year.

Prorail is also working on main line passenger trials of ATO over ETCS Level 2 at GoA 2. In March, the infrastructure manager signed an agreement with the province of Groningen and Arriva, the operator of services on the non-electrified Groningen - Zuidhorn line where trials on services without passengers are set to take place before the end of the year. Critically, the line includes level crossings making it a key test site for obstacle detection systems.

“Through the ATO trial programme, we are keen to become a leader in this technology and for the Netherlands to become a test centre for rail innovation,” Prorail says.