Recent autonomous vehicle (AV) fatalities in the US have raised fresh concerns about safety and viability.
But the reality is that AVs are already here and the industry – and many governments – are adamant they will eventually make motor vehicles dramatically safer than ever.
A Gallup survey from the US this year showed 53 per cent of Americans believe fully automated, driverless cars will be in common use in the US in the next 10 years. Tellingly, that survey was held after the twin tragedies of an Uber test car that hit and killed a 49-year-old pedestrian and a Tesla, using its autopilot function, which killed the driver when it hit a roadside barrier.
However, there is no doubt the crashes have raised questions like: Can we really trust a computer to take our families safely from A to B? What are the challenges to safe autonomous mobility?
Few doubt autonomous vehicles will ultimately make our roads safer, drastically reducing the number of fatalities in the long term. Computers don't get drunk, they don't get distracted and they don't speed – the three main causes of road fatalities in New Zealand.
There are also compelling economic benefits for reducing road accidents. In New Zealand, the social cost of road accidents in 2016 (according to Ministry of Transport data) about $4.17 billion a year. If AVs can reduce that by 50, 70 or even 90 per cent, that is a massive economic benefit as well as averting an immense amount of pain and suffering.
AVs will also allow an ageing population to stay mobile for longer; they can help overcome the two main problems of old age - weak eyesight and slower reaction times.
An autonomous traffic system (even if only partially autonomous) is much more efficient. AVs can "platoon" – closely following behind each other in a continuous stream - using less road space, saving fuel, reducing congestion and improving traffic flow.
So with all these benefits, what sort of vehicle will emerge from the billions of dollars of research that car companies like Toyota are putting into AVs?
Some are working towards completely eliminating the driver. Toyota's strategy is to develop advanced elements such as merging, high speed lane changing and platooning to complement driver skills.
"When we introduce our first semi-autonomous vehicle in 2020, it will still have a steering wheel," says Alistair Davis, CEO of Toyota New Zealand. "That's important because it means the driver will still have control."
He says both strategies are legitimate and can be regarded "as a chauffeur versus a guardian angel. A guardian angel uses as much of the driver's skills as possible, maintaining the fun of driving if the driver chooses."
However, the chauffeur mode makes no use of the driver's ability; the vehicle is simply a transport vessel. From a safety perspective, the "chauffeur" mode is an all or nothing approach with risks along the way until full automation is achieved.
"With the guardian angel approach, the focus is on continuous improvement, reducing lives lost as soon as possible," Davis says.
"Much of the autonomous technology that will feature in these vehicles, like active radar cruise control, lane departure warning, blind spot monitors, pre-crash safety systems and autonomous braking, are already in many Toyota and Lexus models.
"Customers can already experience some of the features and benefits of a guardian angel AV when the car warns the driver of approaching danger and perhaps even begins to apply the brakes if it senses an accident is imminent."
The focus on safety can also be seen with Toyota's wariness over what is known as Level 3 of the clearly defined stages most automotive companies are using to develop AVs.
Level 1 is when some driving assist features are included. Level 2 is when the vehicle has automated functions like acceleration and steering, but the driver must remain engaged and monitor the environment at all times.
Level 3 is known as conditional automation - a driver is a necessity but is not required to monitor the environment, though must be ready to take control of the vehicle at all times with notice. Level 4 is when the vehicle is capable of performing all driving functions under certain conditions, with the driver able to control the vehicle. Level 5 is when the vehicle is capable of performing all driving functions under all conditions. The driver may have the option to control the vehicle.
While Level 3 is a significant leap forward, it is also the most challenging and potentially dangerous; it passes control back and forth between driver and computer. A number of automakers, including Toyota, have said they may look to avoid Level 3 altogether as the risks are too great, only releasing the technology when Level 4 or higher can be fully tested and proven.
In preparing for full autonomy, Toyota has set themselves a target of a trillion miles of test driving. Davis says: "The more testing a vehicle has, the more data is inputted into its artificially intelligent brain. Over time this means the car can make better decisions, having been exposed to billions, even trillions of different scenarios over its training period."
But it's also not just up to the car manufacturers; a safe autonomous future also depends on a massive infrastructure task ahead. Our roads need to be ready – for example, the nation's bridges are often narrower than the road, meaning a considerable obstacle for AVs which expect predictable road layouts.
In New Zealand, that could mean we are 20 years away from being able to reap the benefits of full autonomy.