A Google self-driving car sideswiped a bus this month, the first time one of the company's autonomous cars has been at fault in an incident in four years of testing. Although noone was injured, the accident has sparked questions about liability for driverless vehicles - who's at fault when noone's driving?

The brains behind the wheel - how driverless cars work

Driverless cars were science fiction in the ‘60s, crime fighters in the ‘80s and cute little bubble cars drifting around Silicon Valley in 2012. Components of autonomous systems have made their way into consumer vehicles already, and major technology companies have completed significant tests.

A driverless car requires three key components: a GPS, sensors and an interface to the mechanical aspects of the car. The GPS isn’t that different from the one found in your smart phone; it takes positional data, checks the map and plots out the best route. The shortcoming of the GPS is that it can’t tell a car to stay in a lane, or recognise a detour sign. This is where sensors come in, with a combination of radar, cameras and lasers. The camera provides a visual data set for the car (see this video for an example) allowing it to differentiate between a stop sign and a traffic light. The radar provides object information in ‘blind’ conditions, like night-time or snow storms, and the laser creates an omnidirectional 3D map of the world. This lets the car know how far away things are. All this information is managed by a processor running algorithms that work out where lines are, and the difference between a pedestrian and a plastic bag. It can then tell the car when to start, stop or adjust course to reach its destination.

An accident waiting to happen?

All this might sound a bit scary, but testing has been underway since 2012, with Google logging more than 2 million autonomous kilometers. In this time there have been 17 accidents but the irony is that most were caused by human error, with only one instance so far instance where the self-driving car was at fault. Robots are a lot better at following rules than humans, it seems.

This testing has now hit Australian shores, with the first autonomous car driven in Adelaide on a blocked road, in partnership with Volvo. The test was limited to adaptive cruise control, which calculates driving speed based on the car in front and increases speed when the road cleared. Sensors were used to stay in the designated lane.

While this may seem underwhelming, it’s an important step to developing the regulatory framework for driverless cars. Assessing the applications and limitations of the technology are vital to developing a new framework. This will undoubtedly coincide with a shift in the way we access transport in general, evidenced by on demand services such as Uber, which allow a consumer to order a driver in-app. Combine a service like this with a driverless car and a whole new model of transport could emerge, so it’s clear a new model for regulation will have to be developed.

Who’s to blame when no one is driving?

A big question that needs to be answered is liability: when in an autonomous vehicle, should the car owner be held liable for damages, or should the manufacturer? If the manufacturer is deemed responsible, is their duty of care to the driver, or surrounding pedestrians and property?

The duty of care issues surrounding self-driving cars are complex and many are still to be answered. For example, in a scenario where an autonomous car must either hit a pedestrian or swerve to avoid them and risk a crash that injures its driver, who should it be programmed to protect? While questions like this can be sobering, it’s predicted by McKinsey and Company that 90% of all accidents could be prevented by driverless vehicles.

Regarding liability, it's expected that for the most part drivers will not be held responsible. Recently Volvo announced that it would be the first manufacturer to accept liability for accidents caused by their self-driven cars, and other manufactures are expected to follow. This can be attributed to the fact that is more likely to be the programming or sensors that cause an accident. It's great news for motorists, as it's anticipated the introduction of fully autonomous driving could cut premiums by up to 75%.

At Gallagher Bassett, we understand that disruptive change can significantly impact our customers. Follow us on LinkedIn or twitter to keep up to date with disruptive technologies, and how they impact your business and your premiums.