Google Cars and Law

This blog was first published at Leiden Law Blog under the title "The blind at the steering wheel"

Last Tuesday US senators had the opportunity to listen to a group of scientists and US tech giant representatives. The director of the self-driving car at Google X, dr. Chris Urmson, gave testimony on the present stage in the process of developing a self-driving vehicle and lobbied for new legislation that would facilitate innovation and better fit the evolving market.

In particular, Urmson made an appeal to the Congress for immediate federal intervention to enable additional testing, which would accelerate the roll out of autonomous cars on the market. At the moment Google is testing its self-driving cars in four American states, California, Austin, Texas, Kirkland and Washington, but once the product is launched it has to be able to move across the whole country and this is when major problems may arise – state legislation is highly disharmonized and often self-driving car-unfriendly. If this status quo is maintained the entire rollout plan may be delayed.

Such a delay could be devastating, as US companies are not the only ones trying to bring driverless cars onto the market. China, Australia, Japan and some EU member states have also been working on similar projects and are equally determined to make self-driving cars a reality in the near future.

Great opportunity for the blind and disabled

Driverless cars as a category refers to a wide range of different technologies. The ultimate stage is what Google calls an autonomous car. Autonomous cars are a hybrid between vehicles and computers as they are operated by a complex computer system consisting of cameras, laser sensors, GPS software, and a multitude of other mechanisms that create a 3-D image of the world around the vehicle. These sophisticated vehicles are undoubtedly a great economic opportunity; preliminary research has shown that they can significantly increase road safety, help achieve significant fuel and time savings, and enable disadvantaged consumers such as blind and disabled persons to become more independent. Notwithstanding the big gains, heated discussions have taken place, contesting the level of safety, cyber security and privacy protection of data generated by the vehicles, liability and, last but not least, the ethics of allowing an algorithm to take critical, sometimes life-and-death decisions.

Looking for the least cost avoider

Safety has been a number one issue in the discussion about autonomous cars. Although Google is calming the public with its statistics proving that they have only been involved in 17 minor crashes throughout the entire testing period of 7 years, the danger may be very real.  Car maker Toyota recently settled a class action lawsuit stemming from several instances in which autonomous acceleration systems in certain models malfunctioned, causing the cars to rapidly and uncontrollably accelerate and crash, resulting in severe personal injuries and property damage.

Ex ante state imposed standards are one way to increase the safety level. A regulatory provision for driver’s license endorsement for the operation of an autonomous vehicle was, for instance, introduced in Nevada (USA) which was the first state to pass a considerably comprehensive law on autonomous cars.

Ex post, the level of safety will be assessed through product liability schemes and tort law rules. A couple of authors have mapped the existing liability provisions and linked them to imaginary cases of autonomous car crashes, for instance to a situation when an empty car sent to pick up a family crashes into another vehicle. Whereas some have claimed the current laws are fit enough to address liability issues, others believe the current regime either fails to adequately protect victims or stunts innovation.

Duffy and Hopkins argue, through the lens of law & economics, that the owner of an autonomous car, as he is the least cost avoider (i.e. the actor who can act to prevent a loss at the lowest cost), should carry the burden of strict liability. “The application of strict liability seems like a heavy burden for the owner to bear; however, it is the best system for dealing with autonomous cars for several reasons. Shifting the costs away from the owner and to the manufacturers would stunt innovation and increase time to market.” But is it truly the owner of the car who avoids the economic cost most easily? Are not manufacturers and merchants better positioned to be strictly liable?

Volvo, the Swedish car manufacturer, seems to agree. In a recent press release they stated: "Volvo will accept full liability whenever one if its cars is in autonomous mode, making it one of the first car makers in the world to make such a promise." Certain other manufacturers such as BMW and Mercedes-Benz later expressed the same view.

Should an algorithm take a life-and-death decision?

If an autonomous vehicle is involved in an accident, this can trigger complex ethical questions. Andrew Moore, the computer science dean at Carnegie-Mellon University acknowledged that there would be situations in which a car knew that it was about to crash and would be planning how to crash. “There will be incredible scrutiny on the engineers who wrote the code to deal with the crash. Was it trying to save it occupant? Was it trying to save someone else?”

This is indeed an important question that should be addressed prudently: are we comfortable with the fact that an algorithm can now take a life-and-death decision?

Europe trying to keep up

The UN Vienna Convention on Road Traffic seem to be a considerable barrier to faster deployment of autonomous car testing models in Europe, as it requires that every moving vehicle or combination of vehicles shall have a driver and that every driver shall, at all times, be able to control his vehicle. The UK, one of the countries that has never ratified the convention, though being its signatory, has used this fact to its advantage and has offered its roads to scientists and industry to carry out pilot testing. While an amendment to the convention that would eliminates this burdensome provision is still not confirmed, Germany, Sweden, the Netherlands and France are trying to catch up. Recently, the first self-driving electric shuttle for use on public roads has taken to the road in the Dutch province of Gelderland.

The European Commission is also determined to keep up with the world’s leading economies. The gear 2030 roadmap promises to provide a coherent EU framework for automated vehicles, based on the experiences of the four member states. The Horizon 2020 research program, in which Leiden University also participates, offers finance for those who are working on models and systems related to autonomous vehicles. The EU scientific and academic community seems to play an important role in the technology innovation race, which will, hopefully, soon result in the deployment of safe and prosperous driver-less vehicles.