Listen to this article
Road safety statistics make for grim reading. Almost 1.3m individuals die in traffic accidents worldwide each year, and up to 50m more suffer non-fatal injuries, according to figures from the World Health Organisation.
The National Highway Traffic Safety Administration, the US transport regulator, says the economic cost of the 33,000 or so crashes in the country each year is almost $250bn, whether in lost productivity, legal and court costs, congestion or emergency services.
But much of this could be eliminated if automakers and tech companies have their way: by taking away control from the people who cause more than 90 per cent of road accidents — human drivers.
Self-driving cars are hailed as a game-changer for health and safety on the roads, with their potential to dramatically reduce accidents. “One of the advantages of autonomous vehicles . . . is increased situation awareness,” says Randy Visintainer, director of autonomous vehicles at Ford. “You have 360-degree awareness between your lidar [remote sensing technology], radar and cameras versus the [human] driver, who can realistically only look in one direction at a time.”
Although highly automated cars are unlikely to be seen outside trial settings until the end of the decade, the incremental strides being made towards that point are already bringing safety benefits. Advanced driver assistance systems (Adas) are helping to reduce accidents. For instance, automatic emergency braking (AEB), which uses radar camera-based technology that automatically brakes to avoid stationary obstacles at low speeds, can cut rear-end collisions by 38 per cent, according to research by Thatcham, the vehicle safety experts. The same technology was found to cut injuries by a third. Only a small fraction of new cars sold come with AEB. But this year 10 carmakers pledged to fit the technology — considered the most important safety development since the seatbelt — as standard in all new models. Yet despite these advances, manufacturers and researchers warn that moving beyond basic Adas features towards full automation is riddled with complications.
Fully driverless cars, for example, will have to interact with conventional vehicles and fallible human drivers for some time. That is proving to be a considerable problem. A recent report by the University of Michigan’s Transport Research Institute found that self-driving cars had a higher crash rate per million miles travelled than conventional vehicles. But the self-driving cars caused none of the accidents recorded.
Similarly, cars equipped by Google, the tech company that is seen as the leader in driverless car research, have been involved in a dozen or so accidents, but largely as a result of being rear-ended by distracted human drivers.
“We’re still trying to understand how autonomous cars interact in a completely autonomous environment as well as how they interact with normal drivers,” says Mr Visintainer. “We still envision that these cars will have to interact with human drivers.”
Further complicating the picture, manufacturers from Google to Tesla to Ford are each at different stages of progress with their technologies, and approaching the challenge in different ways. Ford, for instance, is adding Adas features to achieve a high level of automation, like many automakers keen to build market-ready tech that can boost revenues. Ford is also running a separate project to build a Google-style, fully driverless car.
All of this means that there will probably be grades of automated car, further increasing complexity on the roads and presenting a difficult transport infrastructure challenge to governments and regulators. Volvo, one of the leaders in autonomous driving, says it is also concerned that, once automation has reached a certain level, drivers will be so unused to intervening that it becomes unsafe to ask them to take back control.
“We make this [technology] better and better, we add functionality — we come to a point where the driver almost never has to intervene,” says Erik Coelingh, senior technical leader for safety and driver support technologies at the Swedish automaker.
It takes about five to 10 seconds for a driver to fully engage with his surroundings after having been “out of the loop”, focusing on a task other than driving.
“The intervention of the driver is almost never required. If it happens only once every other day, once every week, I would be really concerned the driver wasn’t ready to take over when needed,” says Mr Coelingh.
That means that manufacturers adding incremental pieces of safety kit will at some stage have to make the leap beyond this problematic grey zone and create cars capable of coping with all traffic situations and weather.
Even if manufacturers reach the fully driverless nirvana, the prospect of a totally crash-free future seems unlikely.
“There will be the opportunity for automation to reduce greatly the number of collisions we see on the roads — but not all of them,” says Nick Reed, senior academy fellow at the UK’s Transport Research Laboratory.
“There’ll still be the pedestrian that walks out at entirely the wrong moment and the laws of physics dictate that at some point that kind of crash will be unavoidable.”