Risk Management

Rail revolution: autonomous trains coming down the line

Oliver Lauxmann, Chief Underwriting Office – Liability, Global Practice Group Leader at AGCS and Thomas Berning, Senior Risk Consultant, Liability, at AGCS

Advances in digital technologies are bringing the mainstream rollout of autonomous or driverless trains on longer routes a step closer. What will this mean for risk managers in the rail sector and their suppliers? We talk to the AGCS experts to find out.   

The one – minute dialogue  

  • Digital technologies are facilitating the roll-out of driverless and autonomous trains beyond urban transit systems to longer-distance networks  
  • Pressures to decarbonize transport have given the rail sector more impetus to innovate and create new efficiencies  
  • The risks of autonomous trains vary according to whether they operate in a closed or open environment  
  • Digitization could affect all aspects of rail travel from back-office admin to real-time operations  

In the remote Pilbara region of north-western Australia, a heavy-haul locomotive sweeps across the plains, its 240 wagons stretching two and a half kilometres behind it towards the horizon. Laden with iron ore, the driverless train is making the 300km journey from mines inland to a port on the Indian Ocean, while a team of controllers from the mining giant Rio Tinto monitors its progress from a control centre in Perth, 1,500km away.  

On the other side of the world in Brazil, the fully automated Line 4 of the Sao Paulo metro system carries a human cargo – sometimes as many as 800,000 people a day – along an 11km route through one of the most populous cities on the planet.  

Both are examples of what innovation has made possible on the world’s railways, and how adaptable technology must be to cater for such varied uses and contexts.   

Many of us think nothing of boarding a driverless transit service at an airport or taking a short hop on an automated metro system. Fully automated trains have been around a surprisingly long time – the first opened in Kobe, Japan, in 1981, and they now operate in over 40 cities around the world, including Copenhagen, Paris, Singapore, Dubai, and London. But Rio Tinto’s AutoHaul fleet, developed with Hitachi Rail, was the first driverless, heavy-haul, long-distance train in operation when it completed its first loaded run in 2018. In many ways it represents the changes that are speeding down the track for rail transport.   

Making autonomous rail a reality  

Governments, operators, research bodies, and industries are rising to the challenge of how to take autonomous trains from research prototypes to operational viability, and from urban, closed systems to the more open environments associated with freight, regional and long-distance rail travel. Collaborative partnerships, many of them international, are starting to bear fruit.

The French national railway SNCF, with industry partners, has announced the launch of two autonomous trains it aims to have running by 2023. One will be a freight train, developed with Alstom, Altran, Hitachi and Apsys, and the other a regional passenger train, with Bombardier, Bosch, SpirOps and Thales.   

Another consortium in Finland, led by rail operator Proxion, is developing an autonomous freight train for short distances in the steel and forestry sectors, also by 2023.   

In China, the world’s fastest driverless long-distance train, the Beijing-Zhangjiakou Railway, opened at the end of 2019 and covers the 174km between those two cities in less than an hour at speeds of up to 350km/h.   

“These are just some of the projects taking rail travel in a whole new direction, thanks to ‘smart’ technologies like artificial intelligence, or AI, robotics, and the Internet of Things,” says Oliver Lauxmann, Chief Underwriting Office – Liability, Global Practice Group Leader at AGCS. “Combined with the capabilities of data analytics, satellites, lidar [laser scanning], radar, and 5G, they offer monitoring and safety systems with wide-reaching benefits.”  

The drive to digitize rail travel has been given extra impetus by global efforts to decarbonize and transition to more sustainable forms of transport, as well as the need to tackle congestion in urban environments.  

In France, the government has announced plans to limit domestic flights that can be covered by train in under two and a half hours, while China is aiming to double the length of its high-speed rail network to an astonishing 70,000km by 2035.  

Rail’s role in the global transport mix is expected to increase in future, whether it’s in intermodal freight, long-distance passenger travel, or in and around the digitized, highly networked urban centres of the future – China alone is building 500 of these ‘smart cities’.  

“Disruptive technologies have affected so many areas of our lives, from our phones to our cars,” says Lauxmann. “They support human activity in many different industries and are highly transferable. We’re seeing increasing interest from transport operators in exploring autonomous solutions and we expect they will ultimately contribute to even higher safety standards in the rail sector by reducing risk exposures, including human error.”  

A glimpse of what’s to come  

The efficiencies offered by advanced technology could touch all areas of rail operations. They include optimized speed, braking and acceleration, which some operators say could reduce energy consumption by up to 30%1, and improved network capacity and reliability. Monitoring systems could spot acts of vandalism or theft before they pose a danger to rolling stock or human life and reduce the need for trackside repairs. Geolocation technology could alert operators to hazards on the network and ensure time is made up after stoppages. Data fused from a range of sensors could identify obstacles and calculate their distance from a train, even in the dark, faster and more accurately than the human eye. Data gathered could also support machine learning for more streamlined operations and predictive maintenance.  

The passenger experience is likely to be heightened, too. Travelers could be alerted to their nearest available seat or where to stand on the platform. Higher capacity and greater reliability could encourage people off the roads and onto the rails and even lead to lower ticket prices.   

Train autonomy could also support beleaguered supply chains, by giving logistics providers the data to optimize routes and maximize capacity, reducing the inefficiencies caused by empty containers.   

De-risking driverless rail   

The list of ‘what could be’ in the future of rail travel is long, but the obstacles facing the mainstream rollout of autonomous long-distance trains present real and tangible risks for operators, manufacturers, and their suppliers. The rail industry is highly regulated, but the regulatory framework surrounding AI and other digital technologies is still evolving. The European Union2 is proposing new rules to manage risks posed to the public by AI, with standards for ‘high risk’ areas such as critical infrastructure likely to be stringent. This new standard could have repercussions around the world, and concerns have been raised that it might stifle innovation in the technology sector, while others say the measures do not go far enough to protect human rights.   

 As well as the perils faced by conventional rail operations, such as hazardous cargo, impact, terrorism and geography, some particular exposures affect autonomous trains.   

“Urban driverless systems in closed environments present different risks from networks in open environments,” says Lauxmann. “An item of clothing caught in the doors of a city light railway, for example, represents a different risk from a passenger train colliding with a vehicle on a level crossing. Rail operational exposures vary widely and there is no one-size-fits-all solution. Whatever software works in a mass-produced car won’t apply to all autonomous trains. The technology needs to be highly customized.  

“The main severity-driven incidents we see in the rail sector concern collisions, either between vehicles or with an object, and derailment. These are usually down to human error, possibly caused by fatigue, a missed signal or an issue with speed. Damage can occur to passengers, the train itself, the track, people nearby or surrounding property. These incidents are thankfully very rare, but they are high-profile.”  

The systems used by autonomous trains need to contend with weather events, objects on the line, animals or people intruding on the tracks, and other unforeseen events. “So these trains must be able to react with agility and accuracy to their environment,” says Thomas Berning, Senior Risk Consultant, Liability, at AGCS. “The systems need to be extremely high-performance in terms of on-board intelligence and the integrity of the infrastructure. With such bespoke technology involved, cyber-security measures must be watertight.”  

Berning adds that he sees a possible shift in liability from the human-error aspect to the reliability of the technology used. “And that’s not just software, but hardware too. Was it properly tested and maintained? Have operators and manufacturers taken the requisite steps to protect those systems from attack? This is important because we’re not talking about legacy technology; it’s all very new. The operator is usually the first to be put under the spotlight in case of an incident, but liability does not start and end with them. Litigation could affect the manufacturer, the technology consultants, the designers, programmers, software suppliers… There are multiple parties involved.”  

There are also ethical concerns to take into account, because transitioning to autonomous trains will require retraining staff, negotiating new ways of working and possible job losses, not to mention addressing public concerns over safety and privacy.   

Any risk assessment needs to take a holistic view, Lauxmann says, and AGCS can draw on its global experience of exposures in this space to explore not only insurance solutions but also technical issues with the Allianz Risk Consulting team. “We’ll draw on internal data, external data, and in-house experts to make recommendations on how best to proceed. We see it as like a partnership with a risk manager or broker. For us, the dialogue is key.”  

Related posts
CyberRisk Management

Increasing levels of commercial crime highlight need for formal risk management processes

ReinsuranceRisk Management

The state of the riot wrap market in South Africa

Risk Management

Flood risk modelling could have prepared us for the KZN floods

Risk Management

SA risk body launches guideline on future of risk management