If some manufacturers are to be believed we are already in the age of the self-driving car.
We might not be swooping around in non-stop streams of perfectly synchronised vehicles, but the likes of Tesla with its Autopilot functionality are already claiming that the car is capable of guiding itself around the world’s complex roadways.
While driver assistance systems are hardly new – cruise control has been around commercially since the 1960s – recent years have seen a sudden jump in their sophistication and abilities.
Now our cars can not only sit at a constant speed but use radar to adapt that speed to maintain a safe gap to the car in front. They can keep themselves in lane thanks to camera systems, use ultrasonic sensors to guide themselves into parking spaces, and slam on the brakes if a pedestrian steps out in front of them.
Impressive as all these features are they at best semi-autonomous – they still rely on the driver paying full attention and being ready to retake control should conditions prove unsuitable for the systems. So how far are we from truly self-driving vehicles where ‘drivers’ can put their feet up and read the paper on their daily commute?
Technologically, most manufacturers say we aren’t far away at all. Tesla, for instance, will already sell you a car with all the hardware for ‘full self-driving capability’. Cars run by Google’s autonomous vehicle arm, Waymo, have completed more than two million miles of testing and Volvo, whose Pilot Assist system already features adaptive cruise control and lane-keeping technology, says it already has a ‘complete production-viable autonomous driving system’.
The key, says Volvo, to making the leap from its assistance system to a fully self-controlling one is the Autonomous Driving Brain, a complex network of sensors, cloud-based positioning systems and intelligent braking and steering technologies.
The self-driving XC90 SUVs it is testing in Gothenburg features an array of camera, radar, laser and ultrasonic sensors plus high-performance GPS tracking to constantly map the road and topography around the car, its position on the road and vehicles and objects around it. The data from these sensors is then processed by a network of computers which feed instructions to the steering, throttle, brakes and more to ensure the car is performing suitably for its environment.
Nissan, too, has recently put autonomous systems to the test. A fleet of self-driving Nissan Leafs used similar collections of sensors to navigate more than 300 miles around London earlier this year, handling junctions, traffic lights and roundabouts without incident.
The next step after cars that can think for themselves is cars that talk to each other and the world around them. By broadcasting data such as their position, speed, braking input and more over wireless networks, autonomous cars will be more aware of their surroundings and able to predict and react to things even the best driver literally can’t see coming.
Again, this car-to-car technology already exists and has been tested in the US. General Motors has said it will fit it to Cadillac models later this year and other manufacturers are sure to follow.
In a similar vein, Audi announced late last year that it was installing vehicle to infrastructure (V2I) systems in select models in the United States. The system currently only receives basic traffic light information, and displays ‘time to green’ details on the car’s dashboard but Audi says there is potential for it to do much more. In the future it may be possible to integrate information from the systems into vehicle start/stop features and navigation systems, and eventually help guide self-driving vehicles to ease congestion.
Technologically, then, we aren’t far from a world where cars could pilot themselves safely along our roadways. Nissan says its cars will be fitted with technology to handle junctions by 2020, Volvo expects its Drive Me system to be commercially available by 2021 and Tesla says its self-driving hardware can be activated at the press of a button.
More of a barrier looks set to be the cost; legal and regulatory restrictions and the attitude of human drivers to relinquishing control.
While Nissan’s Leafs successfully floated around London by themselves, the firm’s European general manager for product planning, Stuart Caligari, estimates that every one of the dozens of sensors on each car costs as much as a standard supermini. The self-driving Volvos, too, are built in a special facility rather than the main production line. And earlier this year MIT Technology Review highlighted not only the huge cost of key laser sensors but supply shortage and concerns over their suitability to be fitted to everyday road cars.
Even on cars currently offering semi-autonomous or driver assistance systems the costs are high. Tesla’s Autopilot is a £5,000 option, its nascent full self-driving capability a further £3,000 and that’s on top of a £63,500 starting price. More basic lane-keeping and adaptive cruise systems often command four-figure sums even on high-end luxury cars.
These costs will inevitably fall as research and demand move ahead but beyond them issues such as who bears liability should a self-driving car be involved in an accident, and how to manage roads with both human- and computer-controlled cars on them will give manufacturers and legislators headaches for years to come.
If and when they can solve these tricky questions there’s still the attitude of drivers. For many enthusiasts the idea of allowing a gearbox to shift cogs for them is outrageous enough without even thinking about relinquishing control over throttle, braking and steering.
But for Anthony Ashbrook, CEO of Machines with Vision, the arguments around safety and convenience will eventually win out.
His company has created a system to 3D map the road surface beneath a car, creating a unique “fingerprint-like” image that can identify the exact piece of road the vehicle is travelling along. He believes that increasingly advanced sensors like his firms’ will eventually bring truly self-driving cars to our roads.
He commented: “We are going to see rapid adoption of ADAS [advanced driver assistance system] technology - it's already much more common in Germany for example - but fully autonomous vehicles will take some time and it's hard to predict.
“There are some technical hurdles but the implications for regulation, insurance and social acceptance are substantial and will take some time to work through.
“I do think we'll see a sudden tipping point, however. When the evidence for the safety, economic and convenience benefits are clearly demonstrated things will change fast.
“What happens when parents lobby to have no manual driving zones around schools and similarly city centres are closed to non-automated vehicles?
“We're going to look back in a generation or two and be staggered by a society that accepted 1.2 million road deaths per year and transportation defining many of our urban environments.”