Text size


Driverless Cars - Problems and Answers

Driverless Cars - Problems and Answers


Following the release from Tesla in June and Google’s recent update, Swansway Group question the automotive sectors controversial topic: Driverless Cars.

How will we solve the problems raised by Driverless cars?

There have been 4 fatalities due to driverless cars to date and a number of more minor incidents, does this mean that the safety element, which has been at the forefront of the push for driverless cars, is now defunct?

And if so, what purpose now for the driverless car?

Firstly these incidents need to be put into context; when Tesla spoke about a fatality that happened in 2016, it commented, “This is the first known fatality in just over 130 million driverless miles; in the US, for standard vehicles, with a driver, there’s a fatality every 94 million miles and worldwide, there’s a fatality approximately every 60 million miles.”

These figures suggest, that driverless cars are still up to twice as safe as those with a human behind the wheel; indeed Google publishes a monthly bulletin of incidents which shows that the majority of incidents are caused, not by the technology, but by some form of human error, just as with the majority of accidents involving regular vehicles.

We’re On the The Road To Automation

Most of the vehicles on our roads are already automated to some degree, automatically performing functions which are so basic we take them for granted; switching on windscreen wipers when the screen senses rain, turning on headlights, even opening and starting the vehicle without using the key fob.

A step further down the line, but already in common use are systems which take over parking from the driver such as Park Assist, Adaptive Cruise Control, Lane-Keep Assit  which are available across various brands and models, such as Audi and Volkswagen.

As more technology becomes available, cars will become more and more automated, with the end result being a vehicle which can drive itself with minimum human input. Hopefully this will make for a safer journey and a much fresher driver turned passenger, at the end of the trip.

What’s the hold-up?

American science newsletter Livescience.com believes there are five key areas the boffins need to tackle to allow automation to reach the next level. And there’s one more we’ve added, which concerns us all.



1. Improved software

Steven Shladover, a researcher specialising in transport technology at a Californian university, says the software controlling driverless vehicles will need to achieve safety levels which are an improvement on those currently achieved by human-controlled vehicles.

This sounds obvious, but translates into something which, he says, will be “amazingly difficult” to achieve with the software that’s currently available. He adds that, when Boeing is designing new aircraft, half of the development costs are swallowed up checking that the software controlling the plane is glitch-free. A tall order and a massive cost.

2. Better maps

Dozens of driverless cars are scooting around the streets of Mountain View, California, where Google’s based. They can do this reasonably safely because Google has developed what the guys at LiveScience.com call “a kind of Street View on steroids”. The virtual maps which the cars use are so minutely detailed that the only extra information they need is the movements of the vehicles around them.

This is fine in one American town of about 78,000 people – that’s roughly the same size as Newcastle under Lyme or Bury, but it clearly shows that storing and processing data for larger, built-up areas will mean giving autonomous cars much more ‘brain power’.

LiveScience.com notes that, to date, Google has mapped only about 2,000 miles (3,220 kilometres) of the four million miles (6.4 million km) of roads in the United States. So, we’re barely off the starting grid in this particular race.

3. More sensitive sensors

A cat’s whiskers are their ‘radar’ and very effective they are too, so much so that we’re some way off developing an equivalent, or replicating anything near that level of sensitivity in a robot.

The sensors are, in effect, the driverless vehicle’s eyes, so they need to be able to distinguish between a situation that’s dangerous, and one where there’s no risk and this, say the boffins, is extremely tricky. Without this level of understanding and sensitivity, passengers in the car are likely to be jolted regularly, as the sensors slam on the brakes, every time they detect something.

4. How do we get all these robots to talk to each other?

This is the most important factor. Introducing driverless cars to our roads will mean those vehicles need human characteristics, such as anticipation and alertness. As human drivers we’re constantly monitoring the vehicles around us, noting if a vehicle is being driven erratically and anticipating if that may cause potential danger.

We do all this and just call it driving; the trick is to teach the driverless car to drive; sounds simple, but it’s the hardest part of the process.

5. Building in the ethics

When faced with two possible scenarios, one of which would potentially be more dangerous than the other, how will a robot know which option to choose? Solving this kind of dilemma requires an autonomous drive system to have a whole new kind of capability. It’s what makes a driverless car completely different from an unmanned, remote-controlled missile; if something goes catastrophically wrong with the missile which causes a life-threatening situation, a human can step in and override the controls - what would happen if that ‘missile’ was an autonomous car?

6. Maybe most importantly, what’s the cost of all this cleverness?

One article put it at between £3,000 and £5,300 for the average car, and although market demand will bring this down, it’s still expected to be around the £2,000 mark come 2035. Then again, if all those safety benefits being touted by the cheerleaders for autonomous cars start to take effect, that’ll be more than offset by the savings which will come from there being fewer deaths and casualties on our roads.

UK government figures put the cost of every road fatality at £1.6million and that’s on 2012 figures, the most recent currently available. It’s precisely because there are such high figures at stake that backers of research into driverless cars say that their work is so vital.

Before driverless vehicles can take to the road they need to be seen as part of the safety solution, and that’s exactly why so many people are spending so much time working on the cars of the future.

Given the amount of research going into making driverless cars as safe as possible, are you convinced they’re the future of motoring; if not, what will it take to win you over?