Mercedes-Benz will prioritise the safety of its vehicles’ drivers and occupants – rather than those outside them -- when its autonomous cars hit the road.
And, in stating so, it has answered the tricky ethical question around artificial intelligence in future cars, which has been swirling around autonomous cars and wouldn’t go away.
Now the German luxury car-maker has unequivocally answered the paradox of the no-win choice between saving its passengers or saving an errant pedestrian -- or 10 pedestrians -- by sacrificing one person in the car.
Mercedes-Benz’s future Level 4 autonomous cars will be programmed to save the people they carry, say its Manager of Driver Assistance Systems, Active Safety and Ratings, Christoph von Hugo.
“If you know you can save at least one person, at least save that one. Save the one in the car,” von Hugo told motoring.com.au in an interview at the Paris motor show.
“If all you know for sure is that one thing, one death, can be prevented then that’s your first priority.
“You could sacrifice the car, but then the people you’ve saved, you don’t know what happens to them after that in situations that are often very complex… So you save the ones you know you can save,” he argued.
While a href="/tesla-upgrades-autopilot-103939" target="_blank">Tesla’s so-called Autopilot system has gained notoriety by being seemingly being unable to achieve even that, it’s clear Mercedes’ systems are actively being engineered to keep their occupants alive.
Mercedes is working on a new autonomous testing facility in Immendingen near the German-Swiss border to validate its reams of simulation data, which in turn has been cross-referenced against the thousands of terrabytes of data it has amassed from each new model development program.
But while von Hugo admits the paradox is valid, the safety exec is convinced that it will become decreasingly important as the advantages of ever-alert autonomous cars become apparent.
“We believe this ethical question won’t be as relevant as people believe today. It will be occur much less often.
“There are situations that today’s driver can’t handle, from physical stand point. It [autonomous technology] will be far better than the average driver.
“This moral question of who to save: 99 per cent of our engineering work is to prevent these situations from happening at all.
“We are working so our cars don’t drive into situations where that could happen and drive away from potential situations where those decisions have to be made at all.”
While Volvo and now Audi have both said they will take full responsibility for any collisions, injuries or even deaths their autonomous vehicles may cause, and BMW has also committed to releasing its first fully autonomous car early next decade, Mercedes-Benz has laid claim to the most autonomous production car now available with its new E-Class.
However, the ethical decision that Benz has now made may well be taken out of its hands, depending on the markets their cars are sold into.
A study released in June this year by Science urged law-makers around the world to give a definitive answer to the problem. It spoke out after the majority of people it surveyed said they ethically preferred autonomous cars to sacrifice their occupants rather than crash into pedestrians.
In a classic piece of human nature, though, the majority of the same 1928 Science research subjects said they wouldn’t buy an autonomous car unless it prioritised their safety ahead of pedestrians or other road users.
The joint statement from researchers at the Toulouse School of Economics for the National Center for Scientific Research, the University of Oregon and the Massachusetts Institute of Technology said, in part: “The algorithms that control [autonomous cars] will need to embed moral principles guiding their decisions in situations of unavoidable harm”.
While other artificial intelligence researchers and autonomous car developers have criticized the study, the researchers have set up a website to gather a broader range of opinions on the ethics of the problem.
Yet research from the Stanford Law School shows that human error accounts for at least 90 per cent of all road crashes, along with lesser problems inefficiencies in fuel consumption and traffic flow.
While humans have physiological barriers to unwavering concentration, autonomous technologies combine radar, LIDAR, sonar, stereo cameras and an array of other sensor technology -- all tireless and all running through algorithms and overlaid onto increasingly accurate street mapping.
Even so, von Hugo insists the first generation Level 4 autonomous technology’s abilities won’t be as important as the way people interact with it.
“With the virtual phase it’s not just about the technology and no malfunctions but the people and how they do things and react.
“We have more than 1000 test people in our driving simulator testing various aspects of highly autonomous driving.
“Do they understand how to switch on and off the autonomous? Is there ‘mode confusion’ about what the car’s doing?
“It’s not about miles. It’s about situations and there are an infinite number of them,” the Mercedes executive stated.