New trials of driverless cars on UK roads begs the question: when can I have one? But with ever more sophisticated “driver aids” we may just see a gradual phasing-out of human control as the robots take over, says Matthew Sparkes
This week the Chancellor announced that driverless cars would be tested in British towns and cities to see how they operate around normal traffic. Some £19m would be available to run trials in Greenwich, Bristol, Coventry and Milton Keynes.
Clearly this is good news: autonomous cars promise to reduce fuel consumption, improve safety and free us from the dreary task of city driving. They could drive closer and faster, safely, cutting down on congestion without the need to dig up more green space for new roads. But how long will we have to wait? Not long, for the trials, they start on January 1 and will last 18 months. For production models we can buy and use, though, is less clear.
We’re accustomed to seeing Google’s impressive driverless prototypes methodically plodding through traffic in the bright Californian sun, but how far are we from being able to walk into a dealership and buy one which can negotiate south London’s endless jams to get us to and from the supermarket while we read a magazine?
Nissan says it will launch driverless models by 2020, and Tesla boss Elon Musk says that by next year his cars will be “90 per cent” capable of autonomous driving. But issues with regulation may hold these back. Laws need to be drafted (Who’s at fault if your car runs someone over? Can you “drive” drunk?) and insurance firms need to calculate premiums. Driverless cars will, at first, probably cost more to insure as they’re an unknown risk. But I suspect that once the data reveals that computers don’t drink and drive, check their text messages at the wheel or try to impress their mates with suicidal manoeuvres, policies will plummet. The truly expensive option in 2030 may be electing to drive yourself.
The truth is that fully autonomous cars may be some way off yet, although we’re likely talking years not decades, but they won’t arrive one day in a puff of smoke with a large advertising campaign. Instead we’ll see the existing “driver’s aids” which are already built-in to our cars slowly increase in capability. One day, before we even realise it, we’ll hardly be driving at all. The shift to autonomous cars has already started, we just haven’t realised it.
One of the easiest problems to overcome when making a driverless car, and therefore the first to be solved, is slowing or stopping when objects are directly ahead: autonomous cruise control is already available on a wide range of production cars.
Surprisingly, it first appeared almost two decades ago in 1995 on the Mitsubishi Diamante, but worked purely by throttle control, not braking. Now high-end cars from almost all manufacturers have systems which can maintain a steady speed, braking and accelerating when necessary as a standard feature – something which can be particularly useful in stop-start traffic.
They work via cameras, radar and lasers, but are essentially a tried-and-tested safety feature with a track record of many billions of miles driving.
Similar, but more last-minute and violent, systems are also widespread on both high and low-end cars, known as crash avoidance systems. These will rapidly brake at the last minute prior to a crash in a bid to avoid or minimise the severity of a crash. Often they will wind up windows, close sunroofs and tension seatbelts at the same time to minimise the risk of injury.
This is all well and good in a straight line, but autonomous cars will also need to steer. The first step towards this can in 2000 when Mercedes Actros lorries started leaving the factory in 2000 with a system from US firm Iteris which made a loud rumbling noise if the vehicle left a lane without an indicator being put on. This, too, has now become an option on almost all trucks sold in Europe.
Just a year after that came the first lane-keeping system was launched by Nissan on the Cima in the Japanese market, which not only alerted the driver, but also gently steered the car back into the lane. Two years later Honda launched a similar system on the Inspire, which, for legal reasons, required the driver’s hands to be on the wheel to work – otherwise there would have been a temptation to let the car take over entirely on the motorway for long drives. Was it capable of doing that? Yes, largely, although it couldn’t react to unforeseen circumstances, but just stay between the white lines. Since then many systems from Mazda, Toyota and Ford among others has been capable of the same.
In many ways, braking to avoid objects directly in front of the car and keeping centred between clearly defined lanes on a motorway were easy problems to solve. The truly hard parts of autonomy are the fuzzy parts, replacing the millions of tiny judgement calls that humans make on every single journey in a car.
This is where the bulk of the effort of highly ambitious, fully autonomous car projects such as Google’s are currently being directed. But, again, these are slowly being solved in production cars. One tiny part is traffic sign recognition: cars must know what the speed limit is on any given part of road. The latest Nissan Qashqai, produced in the north east of England, can already recognise these signs – but will only warn the driver, rather than reacting autonomously. Nonetheless, this is a tiny part of the problem solved, and similar features are available from other manufacturers.
Already the driver is being recognised as the weak link. Driver monitors first appeared on the Lexus models in 2006, with a camera on the steering column which tracks the eyes and warns with flashing lights and sounds if a dangerous situation is detected ahead while the driver is looking elsewhere. Toyota also launched a similar system in 2008 which monitored the eyelids for warning signs that the driver was tired.
We’re all driving autonomous cars, in a way, it’s just a matter of degree.