Last updated March 20, 2018 at 2:58 pm
A woman in Arizona has been killed during an Uber autonomous car test, but that’s unlikely to change the future.
An Uber autonomous vehicle undergoing testing. Credit: Uber
Yesterday, around 3,400 people died on the roads around the world. One of those deaths involved a driverless car.
Today, another 3,400 people will die on the roads. But none of those will involve a driverless car.
An incident on Sunday night in Tempe, Arizona has seen a pedestrian struck and killed by an autonomous vehicle undergoing testing by Uber.
Uber have suspended their autonomous vehicle testing programme while investigations continue.
According to the Tempe Police Department, Elaine Herzberg was struck by a vehicle operating in autonomous mode. She was crossing the road outside of a crosswalk at the time.
The Uber vehicle was under the supervision of a human safety driver when the incident occurred.
“Our hearts go out to the victim’s family,” an Uber spokesperson said in a statement. “We are fully cooperating with local authorities in their investigation of this incident.”
In the meantime, Uber have suspended their autonomous vehicle testing programmes using public roads in Pittsburgh, San Francisco, Toronto and the greater Phoenix area.
It is thought this is the first pedestrian fatality involving an autonomous vehicle.
At the moment we don’t know all the details of this incident, and investigations are continuing. Further information about the incident will surely emerge in the near future.
While a tragedy, and an unfortunate bookmark, it is still worth keeping in mind the thousands of people who die in human-driven vehicles every day, which are no less of a tragedy.
This is why the fact that seemingly every incident involving an autonomous car gets widespread coverage is a disappointment. The promise of autonomous cars in reducing road deaths is no less diminished by this, or any other incident, however the splashing of it across the news will have the effect of diminishing public trust and acceptance of this technology which could slash those thousands of daily deaths which receive little coverage.
For all the sensors and computing power onboard an autonomous vehicle, they’ll still (rarely) be caught out by humans doing unpredictable things. It is unreasonable to expect otherwise, especially in this relatively early stage of development.
There is a limit to how quickly the system, and the car itself, can react if a person steps out in front of the car, or if a car suddenly pulls out of a side street.
For the moment at least, that is an impediment to the success of autonomous cars – software trying to understand people’s intentions.
However, in these cases, the software will still be able to detect and react faster than a human. In the case of the Uber incident the car had a human driver on board able to take control at any time. The fact they didn’t suggests a human would not have handled the situation any better – it may have been an incident that caught both software and driver by surprise.
There is also, of course, the possibility that the sensors did not detect the person in the street from some distance, or react appropriately.
The Australian road rule situation
In Australia, the road laws still put the onus on the person sitting in the driver’s seat as being the “legal driver,” making them responsible for any crashes. The legal driver will also be responsible if the car runs a red light, doesn’t give way, or doesn’t stop at a pedestrian crossing.
The responsibility, in Australia at least, starts and ends with the human.
Of course, that raises questions about when, and whether, the human takes control and overrides the autonomous system. In a developing situation, at what point should the human driver realise that they need to take over, and will they even be able to react and take corrective action in time?
Our laws place Australia 11th out of 20 when it comes to autonomous vehicle readiness, however there is much left to do. According to the NRMA, there are around 700 laws which would need to be changed to make Australia truly autonomous vehicle ready.
This is markedly different from countries such as the Netherlands, which allows trials of autonomous cars without a driver. In addition, the country has 1000 traffic lights capable of communicating with autonomous cars, and a thriving research and development industry.
The first laws allowing autonomous vehicles in Australia was enacted in South Australia in 2016, with other states since following. However, much of those laws deal with allowing testing and development on the roads, and not dealing with many of the consumer issues including insurance, liability, privacy and data sharing, and licensing.
Since the SA laws were enacted, over 20 trials of autonomous cars have been proposed, including highway and freeway, off-road and heavy vehicle trials.
With the death of local vehicle manufacturing from Holden, Ford and Toyota, much of the impetus for Australia to become truly autonomous vehicle ready has been deflated, however. No longer are local manufacturers doing research and development for Australian conditions – leaving international companies such as Mercedes, Volvo and BMW to get their head around some uniquely Australian issues from afar.
And dealing with those issues will be a major impediment to gaining widespread approval and acceptance of autonomous cars. Public acceptance and widespread uptake of autonomous vehicles will only happen when consumers are satisfied with their safety and usefulness in everyday situations.
That proof requires testing and development, which will happen. The systems aren’t perfect yet, but they’re getting there. And they will become safer as more autonomous vehicles share the road with less human drivers, who are vastly more likely to do something unpredictable.
Additionally, the acceptance of autonomous vehicles also needs less hysterical reporting by the media over every single incident involving an autonomous car.
