Waymo CEO John Krafcik unveils a Chrysler Pacifica Minivan equipped with a self-driving system developed by the Alphabet Inc unit at the North American International Auto Show in Detroit, Michigan, U.S., January 8, 2017. REUTERS/Joe White
Waymo chief executive John Krafcik unveils a Chrysler Pacifica Minivan equipped with a self-driving system © Reuters

Google’s self-driving cars will remain unplugged from the internet most of the time to prevent them from being hacked, the chief executive of Waymo, its driverless vehicle programme, has revealed.

“Our cars communicate with the outside world only when they need to, so there isn’t a continuous line that’s able to be hacked, going into the car,” John Krafcik told the Financial Times in an interview in Detroit.

“When we say that our cars are autonomous, it’s not just that there’s not a human driver, but also that there is not a continuous cloud connection to the car,” he said, adding that cyber security was “something that we take very very seriously”.

Several large carmakers including Fiat Chrysler and Nissan have fallen prey to cyber attacks, and the risk of hacking will only increase as vehicles contain ever more connected technology.

Because components in a vehicle communicate with each other across a central system, there is a potential for hackers to gain access to a car through one channel, such as an internet connection, and then once inside the car, take control of critical functions such as steering and braking.

In the case of a cyber attack on a Jeep in 2015, hackers gained access to the vehicle’s main functions after infiltrating the car’s connected radio.

Self-driving cars have the ability to communicate with other vehicles to prevent crashes, as well as with parts of the road infrastructure such as traffic lights so they can tell whether they are red or green.

They also send information to the cloud and receive location and other data that help the car to know where it is.

However, Mr Krafcik said the Waymo-fitted Chrysler vehicles “can go for long periods of time without any connection” because all of the driving capabilities are on board the car itself, only opening a link to the cloud to send or receive information such as traffic reports, before closing off the connection again.

This approach is in stark contrast to that of other companies developing driverless technology, such as the collaboration between BMW, Intel and Mobileye.

BMW has said that 5G networks are essential for making driverless cars work because they will need to transmit data constantly and communicate with other vehicles.

Chip designers such as Intel and Qualcomm have launched 5G chips that they say may be used in autonomous vehicles.

But Mr Krafcik said the Waymo cars were not built to be reliant on an external communications network, making them more reliable. “Think about it, if the system goes down and it relies on it, what does it do? Our car is a truly self driving car, it doesn’t require infrastructure,” he said.

Waymo is partnering with Fiat Chrysler to test a fleet of self-driving minivans using the Chrysler Pacifica.

The company announced on Sunday that it would build in-house all of the sensors required to drive the car, in a move it claimed cut costs by 90 per cent.

Mr Krafcik said there was more potential for cost savings once Google begins making the sensors en masse, although he admitted the company might outsource the manufacturing of the components to a third-party supplier.

Get alerts on Driverless vehicles when a new story is published

Copyright The Financial Times Limited 2019. All rights reserved.
Reuse this content (opens in new window)

Follow the topics in this article