30 Aug 2018
How driver profiles could improve AV safety, explained by Mojio’s Kenny Hawk to Louis Bedigian.
Many fear that human drivers and autonomous vehicles will not perform well on the same road. How will they communicate with each other? How will an AV react if a human behaves erratically or vice versa? These are very important questions but Kenny Hawk, CEO of Mojio, thinks there is a way around the potential danger of mixing man and machine.
“Take a step back and think about what they could decipher right now,” said Hawk. “They could have driver profiles. Driver A is a rapid accelerator, brakes hard and changes lanes frequently, so the autonomous vehicle might take a little more precaution when it’s driving next to him. When it’s driving next to grandma, who always drives 55, gently applies her brakes and always stays in the slow lane, that’s a different driving pattern. Thus, without any human communication, you can know a little bit more about the vehicles around you.”
Humans can’t react as quickly
By comparison, today’s drivers have no way of knowing if the driver next to them is about to fall asleep, brake suddenly or cut them off. There are no insights whatsoever and, if there were, human drivers wouldn’t be able to retrieve and act on the data as quickly as a machine. This gives AVs a potential advantage in road safety. “If you think about robo-taxis, these things are going to be driving close to 24 hours a day, seven days a week,” said Hawk. “Versus the cars of today, which are typically only running 5% to 10% of the time. They have to become a lot more reliable with predicting patterns ahead of time.”
Driving patterns are just one aspect, however. To ensure a greater level of safety, AVs must also be able to predict maintenance issues before they occur. “Knowing what needs to be repaired before you have a failure is even more important when you’re in an autonomous environment,” Hawk added. “Those technologies have been around in the aircraft industry for the last 20 years. They’re doing all kinds of analysis – stress analysis, heat analysis, vibration analysis. Even electrical signal patterns coming off the spinning motors and actuators to know ahead of time, okay, this part is about to fail. In a semi-autonomous world for vehicles, we want to have the same kind of thing.”
Hawk said that while normal driving patterns are “relatively easy” to analyze, it could be much more difficult for an AV to respond when a tire blows out. In this scenario, the human driver, who can think and adapt more quickly to random occurrences, might have the advantage.
Hands-free driving isn’t free
Google was one of the first companies to toy with the idea of eliminating pedals and steering wheels from automobiles but it certainly won’t be the last. General Motors has also introduced a wheel and pedal-free concept for ride-hailing services.
Hawk is intrigued by this idea. He said the data shows that technology is better than humans (or will be once it’s perfected). The Uber accident tells another story. If a few more deadly accidents were to occur, that could make it very difficult for automakers to ditch their traditional controls.
“Can you sacrifice a few hundred people to save a few thousand people down the line?” Hawk questioned. “The technology is better than humans. Humans are not fast enough. When you look at the human reaction time for 150kph (93mph), you’re talking many meters before you can react. Many meters could be life or death. Whereas, you look at 5G technology and the speed of processors, robots can make faster decisions better than we can. I think the technology is almost ready but it’s the ethical dilemma of: how do you trade safety versus lost lives owing to technology?”
It’s worth remembering that consumers were also very concerned about the safety of seatbelts, anti-lock brakes and airbags. For the latter, drivers worried that an airbag would deploy randomly while a car is in motion. Hawk said the technology eventually got to a point where consumers were willing to accept that unlikely risk.
AVs are potentially more dangerous, however, which adds to the concern. Hawk does not believe this will be a long-term issue. “When ABS first came out, people were very worried, ‘How is it going to brake as well as I can?’” he said. “They’re thinking that if they jam on the brake, that is the best possible way to brake. Of course, then you start skidding and then you’re not really braking anymore. So think about these other technologies – it took time to perfect them. They launched before they were completely perfect. If you wait for complete perfection, it will never launch. If you look at the adoption curve of each of those technologies, you get an idea of the eventual adoption curve of Level 3 and Level 4 autonomy.”