Just how smart will driverless cars need to be – and how can we ensure they’re safe?

1 min read

A couple of years ago, I wrote that it will be 'some time' before we see driverless cars on the street. Needless to say, I have been proven wrong; Google has been testing its cars on a range of streets in the US for some time, while InnovateUK started tests at four UK sites earlier this year. Already, we heard about the first accidents involving such cars, but it seems it's the 'other driver' who seems to be at fault; at least in the case of Google's cars.

The market for so called intelligent mobility is predicted to be huge – according to the Government, it could reach £900billion in just 10 years. Now, it seems that it's a matter of 'when' driverless cars hit the UK's streets, rather than 'if'. So it's no surprise to hear that British companies are being exhorted to take advantage of the opportunities and the Government is attempting to kick start the activity by putting up £20million. It has also published a code of practice for testing such vehicles.

Topics that are likely to be funded by the Government – participants will also need to put up at least as much of their own cash – include safety, reliability, communication and how driverless vehicles can bring greater independence.

The driverless car will be bristling with technology – sensors, computers, cameras and so on. But operations will be underpinned by software and it seems strange that software is the very last item to be addressed in the 14 page code of practice. The nine lines allocated to the topic essentially say 'don't take a driverless car on the road without first testing the software'.

There's a huge amount of software in a modern car; some say 10million lines of code. How much software might a driverless car require and how can such amounts of code be verified to work in all circumstances? And just how confident might software developers be that the decisions taken in a split second will be the right ones?