Who to blame when hit by a driverless car

The news that an Uber self-driving vehicle has killed a pedestrian in the US has made headlines around the world. It’s a reminder that the era of self-driving cars is fast approaching. But partial or full autonomy raises the question of who is to blame in the case of an accident? In conventional (human-driven) cars, the answer is simple: the driver is responsible because they are in control. When it comes to autonomous vehicles, it isn’t so clear cut.

We propose a blockchain-based framework that uses sensor data to ascertain liability in accidents involving self-driving cars.

Uber has suspended self-driving car tests as authorities gather data about the circumstances surrounding the accident. For partially autonomous vehicles, which still involve human control, assigning liability depends on what action led to the collision and whether it was based on decisions by the driver or the vehicle. For fully autonomous vehicles, the blame can be assigned to, or shared by, one of many parties — including the manufacturer, the service centre and the vehicle owner. Manufacturers could be liable in the case of a design fault, the software provider for buggy system software, or the service centre for inadequate service. On the other hand, negligence liability might fall to the owner for failing to implement a software update from the manufacturer, or with the manufacturer if the accident could have been prevented by a human driver.

In this complex web of potentially responsible parties,

The challenge in this new ecosystem is that some of the potentially liable parties may also have disproportionate control over the sensor data

how can the circumstances surrounding an accident be determined? Fortunately, autonomous vehicles are information-rich platforms with a range of sensors on board that track, monitor and measure everything.

However, the reality is more complicated. The challenge in this new ecosystem is that some of the potentially liable parties may also have disproportionate control over the sensor data. There is a risk that one of these parties may alter the data to steer the liability decision in its favour, using the wireless and USB interfaces that current vehicles already support.

That means we must not only record tamper-free sensor data, but also any interactions with the vehicle.

Blockchain technology can ensure there is untampered evidence of the conditions of an accident to inform decisions about liability. The blockchain framework ensures that the sensor data and records of interactions stored in the ledger cannot be changed without detection. It ensures individual vehicle owners remain anonymous to parties in the operational group. Only the decision partition has access to vehicle owner identities for final liability decisions. This contributes to maintaining user privacy while providing transparent and reliable liability decisions. Using blockchain for trust in sensor data goes beyond driverless cars, extending to smart homes, supply chains and smart grids. The “internet of things” is growing exponentially, and has introduced billions of sensors into our lives, generating unprecedented volumes of data. Blockchain will deliver sensed data we can trust. —The Conversation Raja Jurdak is a Senior Principal Research Scientist at CSIRO. Salil S Kanhere is an Associate Professor in the School of Computer Science and Engineering at UNSW