√Opinion: Why we need to slam the brakes on autonomous cars

It is time to slam the brakes on so-called autonomous car technology, until these complex systems – and humans – can catch up?

In the automotive industry’s race to introduce increasingly advanced driver assistance systems – under the premise of road safety – are we instead creating more dangers on the road and worse drivers?

The rapid roll-out of autonomous vehicle technology – with varying levels of accuracy – is in the spotlight following the tragic hit-and-run crash at a Melbourne tram stop this week involving a Tesla electric car.

The Tesla driver said she was relying on ‘Autopilot’ – the name given by Tesla for its advanced driver assistance software – at the time the vehicle struck the pedestrian, who is now fighting for their life.

Tesla’s autonomous driving assistance systems continue to make headlines as Tesla drivers in the US push the limits of the technology by riding in the back seat, reading a book, or watching a movie while they should be watching the road.

Sadly, there have already been tragic consequences, with numerous deaths and dangerous near-misses in the US involving Tesla cars and autonomous tech. The national road safety authority in the US is now investigating at least nine serious incidents.

Is it Tesla’s fault for using a marketing term that gives owners unrealistic expectations – and which does not accurately reflect the limits of driver assistance technology?

Or is it the fault of owners who don’t understand that, no matter what a car manufacturer says, the driver must remain in control of the vehicle?

Tesla has already been banned by at least one jurisdiction – Germany – from calling the driver assistance system ‘Autopilot’ because, authorities argued, it gave drivers a false sense of what the technology could do.

With more distractions in motor vehicles than ever before, we need motorists to pay more attention behind the wheel – not less attention. If only to keep check on hit-and-miss driver assistance systems.

Earlier this month a Tesla employee was reportedly sacked – and had the ‘Beta’ version of “Full Self Driving” software on his personal Tesla car wiped remotely – after he uploaded a video showing flaws in the technology.

Frankly, it’s not just Tesla pushing the limits. We need all car giants to pump the brakes, so to speak, until the technology can at least be more reliable and more accurate.

Five years ago a leading Mercedes-Benz safety engineer confided in me that the automotive industry is pushing too far and too fast with the rollout of driver assistance systems.

When I shared my concerns with him in a quiet moment with no minder around, he admitted he too had concerns about the limits – and the public expectations – of driver assistance systems on the road to fully autonomous cars.

In my experience – more than 20 years writing about the auto industry and reviewing cars – such systems only work half the time. And it’s clear to me motorists around the world need time to get up to speed with the advancements and the limitations.

It has been such a long and winding road to get to this point, we’ve overlooked how autonomous technology has been fast-tracked over the past decade or so.

We’ve actually been witnessing the rollout of autonomous vehicle technology since rudimentary systems began taking control away from drivers in the late 1950s.

Until the past decade or so, the changes in vehicle driver assistance systems had been so gradual we barely noticed them sneaking up on us. Now we’re struggling to keep up.

Consider this: cruise control (which allows the driver to set their vehicle’s speed and take their foot off the accelerator pedal, but remain near the brake pedal) was experimented with in the 1950s but didn’t become a more common feature until the 1980s.

Back then, the idea of taking one’s foot away from the accelerator pedal must have seemed outrageous.

Today, cruise control on modern cars is a given. However, some motorists after all this time are still nervous about its use – more than half a century after the technology was invented.

Since the early 2000s, thanks to the adoption of radar sensors we have had access to radar cruise control which maintains a safe gap between you and the car ahead, and matches the speed accordingly.

Some radar cruise control systems can even operate in stop-start traffic.

Over the past decade or so, electric power steering systems – combined with tiny cameras to ‘read’ road markings – created lane-wander warnings, or full-blown lane-keeping assistance.

When the camera detects the car wandering from its lane, it can automatically apply subtle brake pressure and/or adjust the steering to keep you in line.

But such systems are still in their infancy and don’t work accurately often enough to be relied upon. Indeed, in some instances, I have experienced this technology steer me towards danger – not away from it.

Which is why I reckon it’s time the auto industry took a breath so the motoring public can get used to all the additional technology introduced over the past decade.

Not everyone is exposed to new-car technology as often as road safety experts, automotive engineers, car company executives and motoring journalists. We still have some people astounded by rearview cameras. They’ve been common for almost 20 years.

The people making the decisions about – and pushing the boundaries of – driver assistance systems already have a head start on the general motoring public.

It’s time the automotive industry and policymakers let everyone catch up before making the next giant leap forward, or else the motoring masses risk being left behind.

In response to ever-increasing safety mandates – as well as executive egos, and fierce competition among automotive giants – the car industry has taken more control away from the driver over the past 10 years than it has done over the previous 70.

Driver assistance aids are being sold to us under the guise of road safety. And I am sure one day cars will be safer than humans. But we are not there yet. In fact, we are decades – not years – away from taking proper control away from the driver.

With autonomous cars, we need much more than sensors, cameras, and radars – and multiple back-ups for those sensors, cameras and radars – and the super-fast processing speed that enables such technology to react quicker than a human.

Engineers also need to recreate what a human brain would do in life-and-death scenarios – as well as the daily grind.

For example, sensors, cameras, and radars can see large objects such as other vehicles and pedestrians – in perfect conditions – but they can’t detect if the driver of another vehicle (or the pedestrian) has spotted the oncoming danger, or is facing the other way.

I test drive between 200 and 250 cars a year across all automotive brands on a mix of familiar and unfamiliar roads. And I can safely say, in my experience, no car company has yet made a faultless autonomous safety system, or driver assistance package.

On balance, such systems work about half the time – and some are better engineered and better calibrated than others.

The poor performers are really poor. Some cars start to brake heavily when approaching steel fences by the side of the road on a bend (mistakenly thinking it’s another car), some cars don’t brake when they ought to.

Some lane-keeping systems work gently and smoothly within the lane markings, some zig-zag down the roadway adding to the driver’s frustration – at times steering them into danger rather than saving them from it.

For example, the lane-keeping assistance technology in the latest Isuzu D-Max ute was so bad, it seemed as though Isuzu rushed it into the vehicle without properly testing it. The company has since announced a free update to the software.

The point is: Isuzu’s original lane-keeping software shouldn’t have made it to showrooms in its original guise. Sadly, Isuzu is not an isolated case.

Advancements in technology – and the willingness of some car companies to push the boundaries of the capabilities of their systems – are now starting to cause unforeseen problems.

Too many drivers are placing too much reliance on technology that’s supposed to be a safety net – not the first line of defence.

Safety watchdog ANCAP – the not-for-profit Australasian New Car Assessment Program that independently analyses crash prevention and protection technology in new motor vehicles – has in part unwittingly contributed to the problem.

ANCAP rightly sets a high standard for vehicle safety. Without its mandates on crash protection over the past 30 years or so, car safety would not be where it is today – and our road toll would likely be much higher.

But ANCAP (and its European equivalent) is pushing car companies to better develop crash-avoidance systems – while, oddly, not mandating two of the most useful and accurate safety systems available today: blind-zone warning and rear-cross traffic alert.

Such systems save lives in busy lane-change manoeuvres (where the driver may not have seen a motorcyclist or pedal cyclist) and in driveways and shopping centres when reversing.

And yet these two extremely reliable and helpful safety aids are not mandatory, but a host of other technology with questionable effectiveness in the real world is rewarded in the star-rating point-scoring system.

The car safety benchmark keeps getting raised every two years but, in my opinion, ANCAP and its international affiliates need to give the car industry – and humans – some time to catch up. Please.

Have a look around at just how far car companies – and customers – are pushing with advanced driver assistance systems, and the risks they are taking.

While the Tesla hit-and-run at a Melbourne tram stop is being investigated by police, we won’t talk about this case specifically.

However, in the 24 hours after that incident, phones across the car industry lit up wondering what the consequences will be.

Is it the driver’s fault? Is it the car’s fault? Each side might blame the other. Police investigators have some work to do.

This incident has the potential to become a landmark case for autonomous car technology in Australia – and perhaps a benchmark globally.

As of today, highly trained police crash safety investigators across Australia do not have a box to tick – or an avenue to explore – when it comes to finding out if or how driver assistance technology either didn’t do its job, or was to blame for a crash.

What police investigators do have – by law – is access to any onboard data and cameras leading up to and immediately after a crash involving a serious injury or death.

“Car companies are terrified by all of this,” said one automotive industry veteran. “Even with all the disclaimers in the world they’re worried they’ll be on the hook if someone dies in one of their cars while it is being used in some form of an autonomous mode. The question is: did the driver place too much reliance on the technology, or is it the car’s fault?”

For now, the law is simple: the driver must remain in proper control of the vehicle at all times, even when driver assistance systems are active or dormant.

For the time being, this is the way it must be because, given all the advanced driver aids – and near misses I have experienced with the technology – we are a long way from absolving ourselves from the responsibility of driving.

As an interim measure, we need policymakers to have stricter requirements on the names and descriptions given for this emerging technology.

Until then, there needs to be more emphasis from the auto industry and policymakers on the responsibilities of the driver – and less marketing razzle-dazzle.

It’s time to stop car makers from using names or descriptions that may exaggerate the level of driver assistance or so-called autonomy offered in their vehicles – and no more demonstrations with hands off the wheel or eyes off the road.

The post Opinion: Why we need to slam the brakes on autonomous cars appeared first on Drive.