July 1, 2016

expert reaction to tesla crash

Expert reaction to the news story that a driver of a Tesla car died in Florida in May after colliding with a lorry.

Prof Alan Winfield, Bristol Robotics Laboratory and Director of Science Communication Unit at the University of the West of England, said:

“This accident tragically illustrates that driverless car technology is still a work in progress. Of course we may never know whether the driver was paying attention at the time of the accident or – even if he was – whether he could have intervened to avoid the crash. My view is that an autopilot that requires that the driver is paying attention and ready to take over in a split second is the wrong approach. It is inevitable that a driver’s attention will wane if they have nothing to do. I also believe it irresponsible of manufacturers to make unregulated autopilot software available for drivers to try out on public roads.”

Prof Will Stewart, a Fellow of the Royal Academy of Engineering, said:

“Autonomous driving looks to be very safe, but not completely so.  But a key difference here is recording; the car records everything that goes on around it, in far more detail than does an aircraft black box recorder for example, so we will know what happened. Thus all cars can learn from this mistake in future, not just the one involved in this particular accident.”

Sahar Danesh, IET Principal Policy Advisor for Transport, said:

“This appears to be a tragic accident but it should not discourage the advancement of driverless cars on our roads.

“Autonomous driving until now has had a strong safety record. One of the advantages autonomous vehicles have over traditional vehicles is that they record everything that goes on around them in detail, so those investigating what happened in the case of the self-driven Tesla will have a lot of information that they can use to improve the future safety of autonomous transport – so that these kinds of accidents can be avoided in the future.

“It is important to remember that driverless vehicles have huge potential to transform the UK’s transport network. In the long term, autonomous cars could improve road safety, reduce congestion and lower emissions.

“There are a series of trials taking place around the world where driverless cars have covered millions of miles without major incident. What we learn from this will prove crucial in ensuring the improved safety and technology of driverless cars.

“Public acceptance and trust are crucial, so these trials must get to grips with the best ways to win over everyone from car manufacturers to consumers to the benefits of driverless cars.

“However, we are unlikely to see fully autonomous vehicles in the very near future but what we will see is increased levels of automation, such as speed and lane control, rather than completely driverless cars. The increased technology in our vehicles will also have a very beneficial effect for older people, allowing them to stay mobile for longer.”

Prof Duc Pham FREng, from the School of Engineering at The University of Birmingham, said:

“The incident shows that there are areas of self-driving technology needing further attention to make the technology safer for people.  This is inevitable with any new technology.  I believe there is a mass market for self-driving cars.  If I were Tesla, I would not be deterred by the incident and would continue every effort to develop safe self-driving cars and make them affordable to more people.”
Prof William Harwin, Professor of Cybernetics at the University of Reading, said:

“Unfortunately, fatal accidents will always happen with new engineering systems. There are any number of examples from the past.

“Sadly, Joshua Brown, the driver who was killed in this Tesla accident, joins others such as William Huskisson, who was the first fatality in a train accident in 1830. Ultimately, Huskisson’s high-profile death did not prevent the widespread adoption of a revolutionary new technology and, I hope, neither will Mr Brown’s.

“It is of course tragic, and there will, and should be, an extensive analysis of how this event has occurred. Tesla should recall all relevant products to at least disable the lane changing feature until this accident can be fully investigated.

“It would appear that the accident resulted from a problem in the performance of the sensors rather than the autonomy of the vehicle. As such, this is a more well-understood problem and there may be a relatively easy solution in terms improving sensor integration, or including additional sensors that are less dependent on light.

“However, the bottom line is that cars are likely to be safer with these automatic features, and ultimately with vehicles that can drive autonomously. Interestingly there has been much dialogue recently among engineers about the issue of liability in autonomous vehicles, and the challenges ahead for engineers, designers and ethicists.”

Prof Slawomir Nasuto, Professor of Cybernetics at the University of Reading, said:

“Accidents like this are tragic, but are sadly inevitable. One of the main advantages of driverless cars is reducing the risk of accidents. But driving heavy metal vehicles around at high speed is by its nature potentially dangerous, and risk can never be eliminated completely.

“Designers of driverless car technology have a difficult paradox to overcome, in that the technology required now needs to be more advanced than later, when driverless cars are more widely adopted. It is much, much harder to design artificial intelligence to replace the role of a driver, operating alongside other human drivers, pedestrians and cyclists, than to replace the whole road system.

“Driverless systems have been operating successfully on trains for decades, although not entirely without incident, yet the interactions on the roads are much more chaotic and complex.”

Prof Nello Cristianini, Professor of Artificial Intelligence at the University of Bristol, said:

“From the explanation issued by Tesla, it seems that there was a very unusual situation that misled the computer vision system, followed by another unusual situation that prevented some of the anti-collision systems from being activated. Modern AI systems are largely statistical operations, and are reliable in most cases, but they may show weaknesses when dealing with entirely unknown scenarios. This is where perhaps traditional reasoning
would give an advantage. The positive side of this technology is that whenever one of their cars encounters a new situation, all the other cars can benefit from that experience, as they all receive software updates.
“As we increasingly rely on autonomous technology, we will see more situations like this, perhaps – one day – involving drones. One way to think about this is that Tesla reports less fatalities in autonomous cars than in the general driving population (1 fatality every 130 million miles against the 1 fatality in 94 million miles for general drivers).”

Declared interests

None received