The road to fully autonomous – or robot – driving is going to be a long one, despite the amount of time and money car companies, infrastructure specialists and legislators are spending on it.

At least a generation away – so some 20 years or so – is the industry consensus despite recent announcements on both sides of the Atlantic that governments want more research to help cut road deaths, serious injury and congestion. Nissan’s view that we could have autonomous driving by the end of this decade is probably optimistic.

Although nearly all the on-board vehicle technology exists today, and most of it will be mandatory on cars over the next couple of years, there are some serious obstacles to overcome including legal issues and concerns about driver behaviour; how do you get a driver to ‘switch back on’ after what could have been some hours on autopilot is how one industry executive explained the concern.

Autonomous lanes for motorways

Autonomous lanes will be introduced on motorways as part of the route to robot driving. GPS and collision avoidance systems, even in their most basic forms, are already very good so the only driver input would be when they need to leave the lane.

These autonomous lanes will probably run at a fuel efficient 60mph with cars travelling close together, perhaps two car lengths apart, to avoid non-autonomous cars cutting in.

Driver Assist Systems

The technology needed for robot driving takes another step forward next year when every car maker will have to fit lane departure warning systems to 50% of each model line. Emergency braking systems will also be mandatory on all vehicles from 2018.

There’s a big prize at stake apart from the human cost in terms of preventing death and serious injury if the car industry and legislators get it right. Morgan Stanley estimates €4.3 trillion could be saved globally each year through preventing accidents. The bank adds in a note: “There will undoubtedly be bumps in the road as well, including the issues of liability, infrastructure, and consumer acceptance. However, none of these issues appear insurmountable.”

Automatic parking

Those bumps will be smoothed out and the first step will be semi-autonomous driving which will also come in stages. The first stage will be parking – pull up outside the supermarket, get out and the car will park itself. When you are ready to go home, you summon the car from an app on your smartphone and the car will come to pick you up, recognising that it is you. Doors and boot will open to allow you to load up your shopping and drive away.

Similarly, at the office cars will park themselves and if you have to go out on a call, you will be able to summon it. This will be very useful at park and ride stations where cars can park themselves while you are getting on the bus into the city and be summoned as you return.

Safety

One of the aims of autonomous driving is, of course, to make travelling by road safer, not just less stressful. Volvo is arguably leading the way with its Vision2020 programme which has a target of no one being killed or seriously injured in a Volvo by 2020.

The company is confident it will get there, noting that no one has been killed in an accident when travelling in an XC90 for the past 11 years in Sweden.

Autonomous driving is a special area of active safety, says Volvo’s head of safety Magnus Olsson, and an extension of what is already available. Aides likes automatic cruise control that keeps you a set distance from the car in front and lane departure warnings already keep drivers away from critical situations, says Olsson.

By 2017, Volvo will have 100 fully autonomous cars travelling on controlled roads in Sweden.

Driver behaviour is one major stumbling block to overcome. How do you get drivers back into the loop if they have been, literally, on cruise control for any length of time?

"Smart" roads

The infrastructure is also a major problem because roads will have to be built to a certain standard, but developments are showing promise. Volvo has just completed a research project using magnets in the roadway to help the car determine its position.

The research has been financed in co-operation with the Swedish Transport Administration (Trafikverket). While established positioning technologies such as GPS and cameras have limitations in certain conditions, road-integrated magnets remain unaffected by physical obstacles and poor weather conditions.

'The magnets create an invisible "railway" that literally paves the way for a positioning inaccuracy of less than one decimetre. We have tested the technology at a variety of speeds and the results so far are promising,' says Jonas Ekmark, preventive safety leader at Volvo.

'Accurate, reliable positioning is a necessary prerequisite for a self-driving car,' explains Mr Ekmark. He adds: 'It is fully possible to implement autonomous vehicles without changes to the present infrastructure. However, this technology adds interesting possibilities, such as complementing road markings with magnets.'

Volvo highlights several benefits of using magnets, including making better use of road space because accurate vehicle positioning could allow lanes to be narrower.

In Volvo’s test, a pattern of round ferrite magnets (measuring 40x15mm) was located 200 mm below the road surface. The test car was equipped with several magnetic field sensors. The research programme was designed to evaluate crucial issues, such as detection range, reliability, durability, cost and the impact on road maintenance.

The next stage will be tests in real-life traffic, which is where the major hurdle still lies. In the same way that the technology exists for hydrogen fuel cells but there are very few refuelling places, all the technology exists for autonomous driving but creating the infrastructure will take time.

Fitting the technology to vehicles is something the industry can do fairly easily; it already has a proven track record of software collaboration through the Autosar development partnership. But getting cross-border agreement in, say, Europe on how to build smart roads is going to be a lot tougher.

Legal issues

The other lengthy process will be deciding the legal issues should an accident happen. Opinion is divided over interpretation of the Vienna Convention on Road Traffic which says, basically, that a driver must always be in control of a vehicle. One argument is that the driver doesn’t have to be a person but could be a robot. But if that’s the case, who is responsible for the robot? The leasing or fleet company which owns the car or the manufacturer or systems supplier of the vehicle? It is going to be a long debate, which may well have to be tested in the courts.

One idea being floated is to establish a trust fund similar to that which helps protect suppliers of childhood vaccines. Vaccines help reduce injury and death – just as autonomous driving should do – but can make people ill – just as there could still be accidents and injury on the roads. In the US a trust fund compensates people who fall ill due to vaccines and at the same time protects the industry from lawsuits.

And when utopia arrives, what happens to the insurance industry? The view is their business model is hardly likely to be affected since accident insurance of some kind will always be needed.