“We are liable for everything the car is doing in autonomous mode… If you are not ready to make such a statement, you shouldn’t try to develop an autonomous system.” So says Håkan Samuelsson, CEO and President of Volvo. A confident statement, also made in similar terms by Google and Mercedes-Benz recently.
The adoption of such a conciliatory attitude could lead to prompt resolution of liability disputes when motor accidents occur, and save insurers substantial sums in settling claims and effecting prompt recovery of losses. But the road to hell (or, in this context, court) is paved with good intentions.
Mr Samuelsson provides the following example of when liability would be accepted: “if the system is causing an accident or over-speeding because it didn’t read a sign in the right way.” But, what if the vehicle had been speeding for some time through a built-up area, and the driver had not noticed? What if the road-sign was poorly maintained by the local council and illegible? In both situations, Volvo would arguably be right in contending that it is not 100% liable, and justified in taking steps to blame another party, in order to protect the reputation of its technology and, ultimately, its sales.
The adequacy of instructions and warnings is likely to be a consideration when determining whether a vehicle, or safety critical component of a vehicle, is defective. A difficulty arises in conveying to a purchaser of an autonomous vehicle what the limitations are, and when they should take over control. Distilling those limitations down into a concise and understandable set of instructions will be challenging, particularly when the same set of instructions will need to be capable of explaining the vehicle’s capabilities and the circumstances in which intervention is required, to the 17 year old novice and the 70 year old veteran equally.
Developers of driverless technology are trialling driving algorithms. An example of a reported problem is where four driverless vehicles approach a roundabout at the same time. All of them submit to the right of way of another, and all grind to a halt. If the vehicles do not communicate with each other, who breaks the deadlock? Will such a cautious driving style be more likely to cause an accident in countries where there is a relatively aggressive driving culture? Will a manufacturer designing a vehicle for sale in Southern Italy, with a behaviour algorithm to match, be liable if an accident occurs in autonomous mode while driving in a country with less aggressive driving behaviour?
Producers trialling the technology and gathering terabytes, indeed petabytes, of data from cameras and sensors for analysis, may benefit from preserving at least some of the more cogent information as evidence, to increase defensibility prospects if, for example, their software algorithms are challenged.
For example, tweaking an algorithm in order to take a slightly more aggressive driving style at roundabouts may lead to less accidents than a more cautious driving style (perhaps because the car behind does not expect the autonomous vehicle in front to stop). In the event that an accident occurs at a roundabout while the vehicle is in autonomous mode and driving in the ‘aggressive’ style, the data gathered during trials may help to demonstrate that the vehicle was ‘safe’, even though it caused the accident. In essence it is a typical risk/benefit analysis.
In a novel area of law, it is sometimes helpful to draw a comparison with more traditional scenarios. In 2009 during the swine flu epidemic, the European Medicines Agency had to decide whether to take more time to evaluate the safety and efficacy of a vaccine, or make the vaccine available sooner. Both actions carried risks. In such circumstances, the legal question is less of the nature of “which is the best choice?” but “can my choice be justified?” Preservation of testing data may help to defend decisions made in respect of vehicle software behaviour algorithms and tweaks to hardware designs.
Volvo’s statement is admirable as it may result in prompt resolution of motor accidents. Rigorous testing should prevent accidents. But, in the wake of the Volkswagen scandal over emissions tests, the question must be asked as to what extent the general public and the insurance industry is willing to place their trust in the performance of safety critical features of vehicles, and the ready resolution of claims.
Written by David Kidman, partner