Now that self-driving vehicles (SDVs) are becoming a reality, people are paying close attention to accident reports involving them. In almost all the cases, it appears that it was an error by the human in the other car that caused the accident. Here is a compilation of dash-cam videos taken from SDVs as they avoid accidents, some of which required quick reactions.
But are the statistics suggesting that SDVs are safer that human-driven vehicles reliable? Peter Hancock says that we should be wary of trusting those statistics because the methodology used to calculate accident rates has problems because we also need to measure non-collision rates as well in order to do a proper comparison and those are not easy to measure.
In a case where there was an accident, the interesting question raised by Srikanth Saripalli is why the SDV did not take the kind of evasive action that a human driver would have.
In Las Vegas, the self-driving shuttle noticed a truck up ahead was backing up, and stopped and waited for it to get out of the shuttle’s way. But the human truck driver didn’t see the shuttle, and kept backing up. As the truck got closer, the shuttle didn’t move – forward or back – so the truck grazed the shuttle’s front bumper.
As a researcher working on autonoous systems for the past decade, I find that this event raises a number of questions: Why didn’t the shuttle honk, or back up to avoid the approaching truck? Was stopping and not moving the safest procedure? If self-driving cars are to make the roads safer, the bigger question is: What should these vehicles do to reduce mishaps?
For reasons such as these, SDV’s will always be a work in progress and we cannot expect the vehicles to have all the answers right out of the gate. After all, the road rules and signs that we now have evolved over time to meet specific needs, as and when problems cropped up that needed addressing. We should expect the same developmental process with SDVs. As Abdesalam Soudi writes, “Driving can involve a range of social signals and unspoken rules, some of which vary by country – even by region or city. How will driverless cars be able to navigate this complexity? Can they ever be programmed to do so?” What will replace the eye contact and other gestures that we use without thinking?
For these and host of other reasons, Jack Barkenbus recommends that that SDV promoters not be too gung-ho. He says that people are not good at rationally evaluating risks and are also exceedingly wary about ceding human control to machines. Too early a roll out that results in highly-publicized accidents involving SDVs and the industry might receive a major public relations setback that will take a long time to recover from.
Another question is who will be the first adopters of the vehicles. What exactly will they be used for and by whom, especially in its early days when the general public would necessarily be apprehensive about adopting this new transportation technology?
The most obvious early adopters are likely to be businesses and governments. This is because although these vehicles have fancy sensing devices, they also depend upon maps to navigate their way and need up-to-date ones. Businesses and bus systems tend to operate on fixed routes and thus are more likely to know about local factors such as lane closings and construction. But how much can such an SDV function without an employee inside? If it is a delivery vehicle then it can get to the destination but what about the last step when the item has to be taken out of the vehicle and placed by the door or in the mailbox? Maybe the solution would be for the vehicle to alert the recipient that it has arrived and get them to come and collect it. But this would require the recipient to be there, which could be inconvenient. In the episode Crocodile in the most recent season of Black Mirror, a pizza company uses an SDV to deliver pizzas and people come outside to the vehicle and pay for it with a credit card and then get the pizza as it emerges from a slot. Having to go outside to get stuff could be seen as a nuisance, defeating part of the purpose of having them delivered. Perhaps the first users would be business-to-business deliveries where there is always someone there to collect stuff. As for buses and shuttle and taxi services, those might be easier to work with since people come to the buses anyway.
Clearly the largest market for SDVs would be regular people who would like the ability to do other things in their cars during their commutes. It might also reduce the need for families to have more than one car if the car could be sent back for others to use at different times of the day. But while this would reduce the numbers of cars sold, the actually number of cars on the road at any given time can increase and so will gas consumption. So that is not good.
As with all new technologies that have the potential to have a mass market, this will be a work in progress for a long time.