Recently we published a long look into some of the big technological changes that need to happen over the next few years, in order for us to get automated vehicles on the road. And that’s a big part of it of course: there’s no automated cars without the AI behind it all; the lasers the sensors, the radar and everything else. Of course, that’s important. But similarly so, we can’t get on this driverless wagon without the human factor being fixed too.
What do I mean by that? I mean things like public opinion, insurance, legislation and the whole system we currently have in place in order to make the roads work for cars that are driven by real people like you and I. How will those aspects of the world need to change in order to make driverless cars a reality on our roads and how will we deal with the issues that crop up from such a system as time goes on?
That’s what I’d like to look into today, to show just exactly what sort of things other than technology that need to be overhauled in order for us to reach that glorious hands-free future.
Insurance is perhaps the most difficult to figure out when it comes to autonomous cars. While accidents in cars with full automation are likely to be few and far between – especially as the number of other piloted vehicles on the road increases – they will happen. Perhaps the algorithm behind it all will have to decide between running into a person or another vehicle, or a wall and a bike. Regardless of how safe the cars are, there will always be some sort of variable that can’t be accounted for: a falling tree, or sudden changes in weather perhaps.
So, at some point, one of these automated cars is going to crash and then you have a quandary: who’s fault is it? Is it the person your vehicle had to swerve to avoid? Is it the vehicle that ran into the back of you? Is it the driver’s fault? What about the company that wrote the algorithm that governed the car’s behaviour, or the individual programmer that wrote that specific line of code?
These are, theoretically, all valid targets. Which would you pick? Of course it will depend on the scenario and thanks to the advanced telematics that will come part and parcel with full automation, we’ll know a lot about the situation, even potentially having laser mapped, 3D images of the scene of the accident as it plays out. But how do you pick?
For now, this is a real “don’t know” category, as really, how can you decide when we’re in the infancy of driverless vehicles? This is one of the key things set to be tested in Bristol next year as part of the Venturer Consortium. It’s not been clear about how it plans to test that at the moment, but presumably it will involve discussing potential outcomes with insurers to see what would work best.
Insurers will of course be keen to find some way to make sure someone needs insurance, as theoretically on roads without regular accidents, people won’t need have much use for it. That has the potential to kill off a big portion of the entire insurance industry, which is worth over £1.8 trillion every year. With that in mind, ministers won’t want to see it go away either, so what they’ll probably do is make it mandatory (for the first generation of automated cars) that all cars need to come with full controls should a driver need to take over and therefore, in the event of an accident, that driver can be held responsible.
There’s all sorts of other reasons that a steering wheel and pedals will be added to piloted vehicles, even if they will never be used, but for insurance purposes, it at least gives everyone somewhere to point their fingers when something goes wrong.
This one is a little hard to quantify, but again it’s something that is going to be investigated by the British teams trialling automated car technology next year. While it’s not necessarily the case that the British public needs to give the go ahead for the technology to be allowed on the roads – the government and regulators will decide that – but it will make things easier. If certain townships don’t like the idea of AI controlled cars or buses, then it could make applying it in a public services capacity much harder.
As it stands, people are quite nonplussed about it all. Most people don’t really know that it exists outside of movies like Minority Report, but those that do are split between those excited by it and a more wait and see attitude. What will be important, is not rushing the technology and leaving it open to criticism in the event of an accident or event that causes damage to life or property. If that happens it will take much longer for the public to come around to it and could leave the UK lagging behind the rest of the world.
Part of people’s introduction to driverless technology will come from their new cars’ safety equipment, which will offer some basic automation in the form of autonomous emergency braking, lane assist and smart cruise control. These will eventually lead to us handing over control of the car during certain aspects of driving like motorway journeys – in convoys – and in traffic jams, two aspect of driving which are the least interesting and frequently boring and frustrating.
It’s this sort of boiling frog analogy, where hopefully because of the slow introduction of certain aspects of the technology, we’ll get used to it and accept each stage rather than freaking out because suddenly overnight Skynet is driving all of our cars. Of course we’ll need to get over certain hurdles, like many of the ones faced by telematics at the moment. Things like privacy, which is potentially going to be infringed if we can’t develop a clever anonymising system or other way of keeping data private.
Although technically more of a technological, or hardware/software focus, it’s worth mentioning security, as it’s something that if our attitude to it doesn’t change, it could give us big problems down the line. As its stands, security is way down on the totem pole for not only laymen, but many IT professionals too. Unfortunately, with connected cars, security becomes massively more important as it doesn’t just lead to your photos being stolen or your iPhone being bricked, it could lead to people dieing.
As it stands, there’s major concerns in the security community that if auto-makers and technology heads that are working within the field of automated vehicles, don’t build in robust security, hackers could potentially spoof a connection and gain access to the vehicle’s systems like brakes, accelerator, steering and more. This might not be such a problem if the user has their own controls which can be used to handle the vehicle, thereby turning off the connectivity, but in one where there isn’t even any way for the ‘driver’ to control it, you could have a big problem.
This is of course, something that needs to be addressed by those with technical know how, but until we start to consider security as one of the major points of software construction, it’s going to be a problem in all fields. It’s a great indicator that this is still the case today, when the likes of Xbox Live and the PlayStation network can be downed by people performing basic DDOS attacks, which while requiring plenty of computers to achieve and do so without getting caught, is an incredibly basic way to attack a system.
Imagine if such a thing happened to a centralised inter-connected car network? If it can happen to a network designed to handle millions of people at once, it could happen to automated cars without difficulty.
This isn’t just a change that needs to take place in the minds of the technically savvy however, but also in the minds of consumers. Wallets drive sales and sales drive development trends. If people began buying more security focused products and informed the companies making them that they take security into account when it comes to making their purchases, many more firms would take it seriously too.
Fortunately this is something we have time to do, as within the next few years trial runs of the technology will be taking place in certain American and British cities (as well as elsewhere in the world) with a specific focus put onto aspects like public opinion and digital security. With public acceptance currently far from stellar, it will be interesting to see if some of the tests showing the safeguarding of the system has an effect on that way of thinking?
Of course, when automated cars start hitting the roads, we aren’t going to suddenly all run out and buy them. For starters, see the commentary above on public acceptance, but it will also be pretty expensive. Google’s pod cars cost up to $250,000 a piece at the moment and even self-drive add-on kits for cars are going to be at least £5,000 in the next couple of years. That means ‘normal’ new cars will be a few thousand pounds more expensive than usual when automated, so not everyone will be able to afford one.
That means that most of us will be driving on the roads, when automated cars begin taking to the roadways in large numbers. Not only do we have to feel envious that we can’t play Candy Crush on our commute like that guy in his driverless car, but we also need to figure out how to drive with them. Piloted cars may be safer in general than one powered by a fallible human, but that doesn’t mean they’ll drive predictably. Perhaps they’ll be ultra safe and we’ll be able to bully them around on the road, but perhaps they’ll have imperfect technology and miss things a human driver wouldn’t.
That probably means that when we spot one of these automated cars we’ll need to take a little more care. Sure we know the driver isn’t drunk or distracted, but until the technology is in its second or third generation, there will likely be teething issues, which means that we’ll need to watch out for certain problematic behaviours. Maybe it doesn’t work so well during wind, rain or snow? Maybe fog presents problems, or high-humidity. We just don’t know yet and that means learning to drive with the AI, before we learn to let it drive for us.
Chances are this is far from an exhaustive list of all the things we’ll need to do in order to accept automated vehicles into our lives, but at least once we do, the piloted systems will do a lot of the work for us. It’s also a good indicator of why the trials set to take place in the UK in the near future are very necessary, as there’s a lot – beyond the technology – that we need to figure out.
Are there any big changes you guys think we need to make (besides the technology) before we fully implement automated cars? Let us know below.