While Elon Musk likes to behave as if driverless cars are just around the corner, the truth is that we’re not going to be cruising around in self-driving vehicles any time soon.
There’s no doubt that self-driving vehicles will transform mobility in the future, it will just be a bit further into the future than everyone hoped. Experts in Silicon Valley and Pittsburgh, two cities at the forefront of autonomous development, predict that fully driverless vehicles - cars that can go anywhere without human assistance - are more than a decade away. Sorry to disappoint.
After years of hype and billions of dollars in investment, companies in the field are realising that their expectations of self-driving vehicles may have been a bit too high. Uber has admitted that it has a long way to go before launching its autonomous fleet. Waymo has yet to open its self-driving taxi service to the public, or move beyond a tiny geo-fenced area in Phoenix, Arizona. Even Tesla has been forced to scale back the hype surrounding autonomy. The company initially promised to have driverless cars on the road in 2017 - a timeline that has stretched as they realise the magnitude of the challenge facing them. These obstacles, while not insurmountable, are proving difficult for researchers to figure out.
The first roadblock to overcome is getting vehicles to operate in diverse weather. There’s a reason testing is done in warm climates like Arizona and California - the sensors that autonomous vehicles rely on are obstructed by snow, rain and fog. In bad weather, the car’s computer has no idea what it’s looking at. It’s like someone who needs glasses driving without them. Software is being developed to solve this problem, but most companies are still trying to figure out how to drive on a clear day, let alone in a blizzard.
Another difficulty is how road markings vary between cities, states and countries. Lane lines and road widths aren’t standardised, so vehicles have to learn how to drive in each different area. For example, your autonomous car may be fine driving around inner-city Brisbane, but become useless among the trams, hook turns and U-turns of Melbourne. You can forget about your annual family road trip too - engineers are yet to figure out how to drive on dusty rural roads or through dense bushland, a huge issue in Australia.
One of the major realisations engineers have made about self-driving cars is that they struggle in similar instances that humans do. Deciding when to turn in front of oncoming traffic without a green arrow can be difficult for any human driver, and autonomous cars have the same problem. The vehicles also find it difficult to make decisions at busy intersections and roundabouts. When it comes to making split-second judgements like stepping on the breaks to avoid hitting a child, or swerving to avoid a kangaroo, self-driving vehicles just aren’t up to the task yet.
Imagine that a driverless car calculates that it should hit a child to save its passenger. Or should the car prioritise pedestrian safety over its human passenger? This isn’t just moral philosophising, it’s a real nightmare for liability and insurance. Between the passenger, the vehicle and the manufacturer there’s a legal gray area around who should be held responsible for an accident. In Australia, the current legislation states that the human passenger is legally deemed the “driver” of an autonomous vehicle, but a self-driving car accident could easily be seen as a design fault that the manufacturer is responsible for, just as Apple is responsible for your faulty iPhone.
When there are accidents involving driverless cars, the public is far less forgiving than they are of human drivers. From the incidents involving Tesla Autopilot to the fatal crash of one of Uber’s driverless cars, the tolerance for mistakes made by self-driving cars is close to zero. Conventional car injuries and deaths happen every day, but people expect near-perfection from autonomous vehicles. Harsh, sure, but people just don’t trust self-driving cars yet. For this to change, companies will have to convince people that driverless vehicles are at least as safe as human drivers.
Companies also have to prove the safety of their vehicles to governments around the world. Australia has started preparations for the arrival of automated vehicles, promising a federal law by the end of 2020, but there’s a mountain of regulations to get through before they can be safely rolled out around the country. Currently ranked 15th on the Autonomous Vehicles Readiness Index, Australia also requires more high-quality road and digital infrastructure before self-driving vehicles can be deployed around the nation. That will only come with major financial investment, something that can be difficult to rely on.
For now, the world is too unpredictable and varied and the robots too expensive and undeveloped for fully autonomous vehicles to compete with human drivers. The good news is that driverless tech is still in its infancy and innovation is happening at a rapid rate. Eventually, these obstacles will be in the rearview mirror as self-driving cars accelerate to new horizons (and onto our roads) - in the meantime we will just have to put up with driving ourselves.