Software in charge – liability and selfdriving cars
Key words: Self-Driving Cars – Software – Private Law Tort Liability – European Union – Harmonisation – Legal Theory
The generalization of car traffic has been one of the most important evolutions in Belgian tort law since the adoption of the Code Napoleon. Society in general and liability rules in particular will undergo another transformation with the introduction of self-driving or autonomous cars. Today, technology has already partly taken over some of the driver’s tasks in controlling the car. Examples thereof are Adaptive Cruise Control (ACC) or Lane Keeping Assistance (LKA). This technology will ultimately be developed to a point at which vehicles are able to take persons and their goods from one place to another without any human interference.
The use of self-driving cars has several benefits. Firstly, transport will become more time efficient. Persons in the vehicle might be able to do other things such as working or even enjoying free time instead of being focused on the driving task. This, as well as a predicted reduction of traffic jams, has a positive influence on the work/life balance. Secondly, self-driving cars have environmental benefits as well because of more ecological driving behavior and the reduction of traffic jams. Thirdly, autonomous cars allow persons that were previously not able to drive a vehicle (e.g. disabled person in need of a wheelchair, elderly or minors) too fully and independently participate in traffic. Fourthly, traffic becomes much safer as self-driving cars decrease the amount of accidents on the road drastically.
Although self-driving cars have many benefits, several obstacles remain and need to be solved as well. Firstly, recent accidents involving Tesla and Google self-driving cars show that the technology still has several flaws and is far from being perfect. Secondly, it remains uncertain whether drivers are already willing to hand down the control of their cars (and thus their safety) to software systems, which they cannot manage. Besides the fact that people do enjoy driving themselves, the distrust towards software taking over the control is eminent. Thirdly, the underlying technology used for self-driving cars is still very costly today. This makes the commercialization of autonomous cars less interesting. Finally, several legal and practical challenges remain. The current stance of Belgian law, for instance, is not adopted to the development and introduction of self-driving cars. The Highway Code (Wegcode) requires that each vehicle has a driver. This requirement might be problematic with the introduction of autonomous vehicles. Moreover, the driver must at all times be able to perform the necessary driving actions and have his vehicle under control. The driver needs to be able to do the necessary man oeuvres and needs to have control over his vehicle. Put differently, the technology used in autonomous vehicles has to allow the driver to intervene whenever it is necessary. If this is not the case, the driver is not allowed to go on the Belgian roads with his vehicle. Additionally, road and traffic regulations might need to be reconsidered in the future because software operates differently than humans. Computers, for example, possibly don’t need traffic lights or signs as they are able to communicate with each other. Thus, they can decide who needs to give way or avoid crossing a roundabout at the same time.
Self-driving cars are no longer a mere futuristic idea and they will be commercialized one day. According to the last predictions, fully autonomous vehicles could be available on the market within five to twenty years. The common use of autonomous cars will have a considerable influence on traffic and on our society in general. Families might, for instance, no longer need a second car but instead rely on cheaper services offered by taxi firms operating autonomous vehicles. Software might over time replace human drivers, which will affect employment in different industries. Moreover, park houses might become redundant with autonomous vehicles as the latter can simply continue driving around and transport other persons, without necessarily having to park. On the long term, autonomous cars might also have an influence on public transport and infrastructures. An extensive rail work for trains or trams would also lose its importance with self-driving cars. The benefits of public transport such as allowing people to relax or do something else than paying attention to the road are also given by autonomous vehicles. Finally, experts predict a decline in the so-called ‘crash economy’ once software is in charge. Garages, lawyers, insurance companies, hospitals or physical therapist would have to bear the consequences of safer traffic.
Although traffic becomes much safer because of car automation, road accidents will not suddenly disappear. Software used in autonomous cars can, for instance, be hacked. This in turn leads to new safety and privacy issues. More importantly, autonomous cars will have to share the way with “normal” none-autonomous cars. There will thus be an interaction between human drivers and software on the road, inevitably leading to accidents. Recent accidents with Google or Tesla self-driving cars show that accidents can indeed happen with such vehicles. More specifically, the question can arise as to who should be held liable in case self-driving cars cause damages. The presentation will try to provide an answer to this question.
Firstly, the presentation gives an overview of the benefits, challenges and problems that go along with the introduction of autonomous vehicles. Secondly, several grounds of liability on which victims can ground their claims to recover losses incurred by an accident with a self-driving car are examined. Attention is given to two particular regimes under Belgian law: fault-based liability of the potential driver (e.g. negligence or violation of a legal duty) and product liability for the manufacturer of the car and/or software (under EU legislation dealing with liability for defective products). Thirdly, we make some proposals in the context of liability and self-driving cars that could be adopted by the EU. More specifically, we argue that the manufacturers of software should be liable when the car causes damages thereby, however, using a risk-utility approach, instead of the common consumer expectation test.
