We Need To Crack This Before We Even Think Of Driverless Cars
We Need To Crack This Before We Even Think Of Driverless Cars
Let's solve the original trolley problem first.
Driverless cars are meant to revolutionise the world. That is, if they ever come to fruition. They are still not reliable enough and incapable of driving safely, let alone resolving the famous Trolley Problem. Well, many are now saying we should solve the original Trolley problem, i.e. self-driving trains, before we even dream of tackling self-driving cars. Self-driving trains exist, but are incredibly rare and limited in their use. When you think about it, this seems rather odd. Trains should be incredibly easy to automate compared to cars. So, why aren’t all trains fully automated? Well, answering this demonstrates just how far away we are from safe self-driving vehicles.
It turns out that London has been trying to convert its famous Tube to driverless trains for well over a decade. But, a document leaked by TfL in 2020 detailed why this switch hasn’t happened: cost. This document spelt out how such a conversion would cost over £7 billion! As such, the authors concluded that introducing unmanned vehicles presented “no additional benefits for the level of investment.” Moreover, upgrading the network in this way would only cut the number of Tube staff by 10–20%, meaning that it won’t mitigate the issues of having workers, such as strikes.
However, London and many other capital cities already have self-driving lines, such as London’s Docklands Light Railway (DLR), which has been driverless since 1987. The DLR isn’t fully automated though, as it has staff onboard to handle customer service, ticket checking, and take control in an emergency, such as a passenger falling between the train and the platform. In fact, many of London’s lines work like this, including the Victoria, Jubilee, Central, Northern, District, Circle, Hammersmith and City, and the Metropolitan Lines.
There are a handful of fully automated train lines out there that do away with the onboard staff. However, these require dramatically changing the platform to have doors between the platform and the train to manage ticket and safety issues. This, in turn, makes them very expensive and, by their own nature, isolates them to the few stops with such weird infrastructure. These lines also require far more infrastructure along the line, such as a dedicated communications network, new signalling systems, trackside components such as transponders, beacons and RFID tags, a centralised Train Control Centre, onboard equipment such as computers, sensors and control units, safety systems such as obstacle detection systems and emergency braking systems, a reliable power supply infrastructure, dedicated maintenance facilities and finally redundancy and backup systems.
But even after all of that, you get a train that is no faster or more efficient than only piloted by human workers. In fact, because of all of these interconnected systems, automated trains can be far less flexible. What’s more, these systems only work for small, slow and isolated train lines, hence why all of these are metro lines or subways. Outside, fast, open network train lines need way more safety features and infrastructure changes to deal with challenges like road crossings, inclement weather, system disruptions, etc. This, in turn, would add such a huge cost that it would be unfeasible.
But, one argument is that these automated metro lines are safer, as there is less chance of human error. But that isn’t wholly true, as these automated systems still fail. For example, in 2012, a driverless train on the Jubilee Line in London had incorrectly given the all-clear for the train to start moving. The staff member onboard noticed that a child had slipped between the train and the platform edge and stopped this action, saving the child’s life. Tales of these system errors, either hurting people or human workers having to step in and correct potentially dangerous operations, are rather common with all of these semi-automated and automated metro lines, and many have happened in the past few years.
Now, a train is a far easier thing to automate than a car. It’s a far more constrained task, with fewer variables and a more controlled, less cluttered environment. If we still can’t fully automate trains without having to dramatically modify their infrastructure, rendering the entire project financially unfeasible, or at the very least, financially challenging to justify, and even after all of that, these trains still have safety issues, why do we think we can fully automate cars successfully? Surely, we need to crack driverless trains first, and surely, the lessons from this will help create better self-driving cars. I know why this approach will never happen, but that tangent is worth an entirely different article that involves references to Who Framed Rodger Rabbit and Musk. So, for now, this is at least food for thought so that every time we see marketing for self-driving cars, we can question what is happening beneath the surface.
Thanks for reading! Content like this doesn’t happen without your support. So, if you want to see more like this, don’t forget to Subscribe and follow me on BlueSky or X and help get the word out by hitting the share button below.
Sources: Future Rail, Verdict, Triptex, Wired