[SCMP Column] Driving towards Driverless Cars

June 06, 2016

Fads come and go. Last week it was 3D Printing that will change the world as we know it. Today it is driverless cars.  Sorry to be an old fashioned curmudgeon, but dream on.

My prediction for the week? Fully automated vehicles capable of driving in every situation will not be here in most of our lifetimes. I know this will outrage the hyperventilating techies who see driverless cars transforming our lives within the decade – some are even talking about 2020 – but I stand my ground. You and I will likely be long dead before such changes happen.

Yes, I know Google cars have achieved lots. And I know many motor manufacturers are pouring significant millions of dollars into the creation of the driverless car. Uber for example just stripped 40 top robotics researchers from Carnegie Mellon University’s highly-regarded robotics department. As David Pogue, anchor columnist for Yahoo Tech, noted in the latest Scientific American: “This is exciting stuff. Self-driving cars, in theory, could eliminate the crashes that kill 1.2 million people every year around the world. Trillions of dollars would never have to be spent on hospital stays and insurance payouts.”

But flick a few pages forward in the same Scientific American June edition, and a more sober prognosis for the driverless car emerges: “Soon electronic chauffeurs will take us wherever we want to go, whenever we want, in complete safety – so long as we do not need to make any left turns across traffic. Changing road surfaces are a problem, so are snow and ice… And in an urban environment where pedestrians are likely to run out in front of a car, we should probably just walk or take the subway.”

In a ruthlesslessly realistic evaluation of our miraculous driverless future, Steven Shladover at the Institute of Transportation Studies at the University of California has a sobering prediction: “Fully automated vehicles capable of driving in every situation will not be here until 2075. Could it happen sooner than that? Certainly. But not by much.” You might not be dead by then, but I almost certainly will be.

Of course, sooner rather than later all depends on what you define as self-driving cars.  And here, Shladover says “increasingly unrealistic expectations” have arisen out of failures to distinguish between “autonomous”, “driverless” and self-driving” cars – and of failures to be exact about exactly what types of environment such vehicles might be able to operate in.

If we are talking about clever cruise control systems, of lane-keeping systems, or what Shladover calls “traffic jam assistants”, then technologies are arguably already in place: “Cars are pretty smart these days. Yet it is an enormous leap from such systems to fully automated driving.”

For this, Shladover and his team have established  a five level “ladder of automation” that starts with driver assistance and partial automation, on to conditional automation, high automation and full automation. We have moved cautiously into Level One, and are paving the way for Level Two, where we will be able to combine automatic speed and steering control – and this is what the techies are getting excited about.

But higher ambitions are going to take years. Amend that: decades. Main barriers are going to be regulators and safety experts who insist that automated driving systems, once unleashed in our communities, will be no less safe than human drivers – which means bettering the current situation where a fatal crash occurs one in every 3.3 million hours of driving, and where crashes resulting in injury occur once every 64,000 hours of driving: “Reaching this level of reliability will require vastly more development than automation enthusiasts want to admit,” says Shladover.

Obstacles are not just technological – they are financial too. He notes that almost half the cost of new commercial or military aircraft is spent on “software verification and validation” – or in simple English, making sure automated systems don’t crash on us. He notes that automated systems managing vehicles moving on roads in close proximity will have to be many times more sophisticated than the systems that keep planes safely apart: “Manufacturers will need to “prove” the safely of a complete automated driving system to the satisfaction of company risk management officers, insurance firms, safety advocates, regulators, and of course potential customers.”
He argues that Level Three automation – automated systems that rely on a human driver to take over control if something goes wrong – will be problematic for decades, because of the problem of speedily “recapturing the attention” of a human being who has spent the past 60 miles watching the scenery, or dozing to sleep.

But prospects are more promising for Level Four automation, where automated systems do everything, but in highly controlled environments. – like automated valet-parking, or vehicles moving in segregated or low speed zones like university campuses or business parks or urban pedestrian zones, or along segregated bus or truck-only lanes. Even these are likely to be a decade or more away, and may only be operable in good weather, and in areas that have been mapped in meticulous detail. And anyway, this is not what the techies are getting excited about.

So I revert to my original curmudgeonly unexciting point: the day is still far, far off when any one of us will be able to step into a car in the garage and instruct our automated chauffeur to drive us across Hong Kong and drop us off in the car park outside our office, or to drop the kids off at school. Until then, those that actually enjoy driving a car may still be able to look forward to the daily gladiatorial challenge of fighting through the traffic gridlock from Admiralty to Central. And those of us that would prefer to be whisked to work in safety and comfort on the MTR might still have chance to fight our corner. Life will go on as normal for much longer than many of our starry-eyed techies want to admit.
[ Back ]