Road danger is a man-made crisis, with human error accounting for more than 90% of traffic accidents worldwide. Autonomous automotive technologies promise a future of fewer crashes and road fatalities. Eventually, zero road fatalities could result from
the advent of fully autonomous cars, which BMW says it can bring to our roads by 2021.
However, the recent fatal traffic accident involving a Tesla Model S that crashed while using its Autopilot feature offers a stark reminder of how far the technology has to go before fully autonomous vehicles arrive. Against a bright spring sky, the car’s sensor system failed to distinguish a large white 18-wheel truck crossing the highway, Tesla said in a statement. The car attempted to drive at full speed under the truck, “with the bottom of the trailer impacting the windshield of the Model S”.
Public perception and acceptance of autonomous cars will be fundamental to their ultimate success, says Ben Gardner, a solicitor at law firm Pinsent Masons who specialises in autonomous cars. “If the technology is not safe, reliable and capable of delivering the societal benefits it promises, then people simply will not want to buy or use them,” he says.
While the Tesla crash has put fresh scrutiny on emerging autonomous technology, the incident was the first in just over 130 million miles where the Autopilot was activated. Among all vehicles in the US, there is a fatality every 94 million miles. Worldwide, there is a fatality every 60 million miles.
Barriers to acceptance
“News of fatal crashes can only be detrimental to the public’s perception and ultimate acceptance of driverless cars,” says Gardner. “However, this news should be contextualised against the backdrop of the fact that road fatalities unfortunately happen every day.”
Arunprasad Nandakumar, team leader of autonomous driving and connected mobility at consultancy Frost & Sullivan, says that the industry must work on many technological developments in the push towards the fully autonomous car. He adds that breakthroughs in hardware, software and service are still required before a more autonomous car is even possible. “The benefits of a lidar system, for example, are obvious, but there are parallel systems that help to improve software by using the existing hardware in the car.”
Nandakumar believes the quantitative benefits of hardware will plateau eventually, and the driving software behind the hardware will eventually lead to better systems of autonomous driving. “For the car to be predictive in the future, a shift in software is the biggest gap that the industry needs to fill beyond just the hardware solutions,” he says.
“For example, technologies that need development are artificial intelligence and digital processing of the data that the camera already captures. The entire aspect of service, including high-definition mapping and many other validation layers, will also be essential in the future.”
While the Tesla fatality drew attention to the increased need for development in autonomous technology, these changes needed to happen anyway, says Nandakumar. “Today’s sensor solutions and the software are technically incapable of becoming a fully autonomous system by themselves. We need to go through those progressions eventually, regardless of what happened,” he says.
Dr Stefan Sommer, chief executive of German automotive supplier ZF Friedrichshafen, agrees. “Self-driving cars will need multiple detection systems including expensive infrared lidar technology if they are to be safe at high speeds,” he says.
Although he refuses to comment on the Tesla crash specifically, in a speech before the incident Sommer had explained that cameras relying on visual signals alone were insufficient for safe autonomous driving at high speed: “For autonomous driving, we will need three core technologies: picture-processing camera technology, short- and long-range radar, and lidar.”
For the industry, the fatality has brought a greater sense of awareness, says Nandakumar. “I think it will force some OEMs and suppliers to re-emphasise validation and testing a bit more before bringing it to the product. They’re already taking a cautious approach and not rushing to market, as they would rather bring about a sustainable solution than the first solution.
“I think the Tesla crash might or might not be progressive to the autonomous driving solution, but it will at least make a change in the way the industry looks at autonomous driving,” he says.
Of course, the Tesla Model S is not a fully autonomous car, and the Autopilot function that was in operation is around level 2 in terms of autonomy – with level 0 being completely driver-controlled and level 5 being fully autonomous. “If the car was completely monitoring the driver,” says Nandakumar, “the accident could have been completely avoided.
“As we progress through the levels and have robust solutions, we will have a decrease in those road accidents. Eliminating the driver out of the equation is the end goal of autonomous driving from a safety perspective, but before then the industry needs to tackle the technology challenges that exist in between.”
Before more levels of autonomy can even be considered on the roads, however, legislative hurdles must be dealt with. The Department for Transport has released a code of practice that those testing autonomous cars will need to comply with. Some of the requirements of the code include having a qualified driver who can intervene if necessary, and the vehicle must be safe, roadworthy and operated in compliance with existing laws. It is also recommended that the vehicle should have successfully completed testing in a controlled environment.
Framework for development
Legislation may not necessarily be a bad thing when it comes to autonomy, says Nandakumar. “Regulation can either hinder the progression of technology, because of the lack of open-mindedness, or it can create a framework where
people know what they’re working towards so they can accelerate developments,” he says.
“We don’t know how autonomous
cars can be monetised and what opportunities exist, but from a technological and commercial perspective regulation gives a framework to help the industry work closer towards that, rather than working towards uncertainty.”
In this case, regulations for autonomy may not be a hindrance as is often suggested in the industry. “There is currently no regulation as to how autonomous cars need to develop beyond just testing.
So I think a proper, open-sourced framework will allow various product developers to share critical data that will accelerate the growth of technology rather than decelerate it,” he says.
Insurance update
The government has recently announced that motor insurance and road traffic laws are to be updated in the UK to account for the expected rise of autonomous vehicles.
The Department for Transport has opened a consultation on measures to support the development and adoption of driverless cars. These measures include plans that might see registered keepers of driverless cars obliged to have insurance cover in place that accounts
for manufacturers’ and any other entities’ product liability, as well as injuries to the ‘not
at fault’ automated vehicle driver, and to passengers and third parties.
Gardner from Pinsent Masons explains that the government is keen to attract businesses to the UK to develop, test and commercialise autonomous car technology, and that overly strict laws could encourage those companies to do this in another country. “The key will be to balance regulation and deregulation so as not to stifle development and testing, while ensuring that the technology can be rolled out safely,” he says.
“However, any accidents caused by driverless vehicles at this stage of their development could undermine consumer acceptance of the technology in the longer term.”
Today, the people who take up the technology are the early adopters, says Nandakumar
at Frost & Sullivan. “We haven’t reached the progressive curve where the majority of consumers would start buying autonomous technology. Early adopters are always risk takers in any product, and are willing to take that uncertainty upon themselves.
“In the short term, therefore, I don’t think incidents such as the Tesla crash will make a big impact on people consuming autonomous driving technology. But in the long term, we will see better solutions in the market that will help the product go beyond just these early adopters into the volume adopters as well.”
Changing risks
The introduction of autonomous vehicle technology should inevitably lead to a fall in the number of collisions on the roads. “The long-term vision of accident-free driving is no longer in the realms of science fiction,” says Gardner. “However, even where the technology is able to prevent 100% of collisions, there will be new risks such as hacking and cyber security to contend with, which could still lead to accidents happening.”
In the medium term, it will be interesting to see how we will cope with a mixture of human-operated and autonomous cars on the roads. “This ‘mixed use’ system will potentially be the most difficult to navigate on the way to full autonomy on our road network,”he says.
The question, therefore, should ultimately be: are fatalities less likely when driving in an autonomous car? Gardner says: “Hopefully this is a question that will be answered as the technology continues to be developed and rigorously tested over the coming years.”
Did you know? Staying in control
More than 65% of motorists want to retain the right to drive even though driverless cars are coming, according to a survey by IAM RoadSmart (Institute of Advanced Motorists). The respondents in this 65% believe that a person should always be in control of a vehicle, with 53% saying that the focus should be on making drivers safer – not just cars. IAM RoadSmart says that, while its members welcome the high-tech advances improving vehicle safety, they also want to maintain their control of a car – even though autonomous technology will be able to do the job for them.