After years of procrastinating, I had my first driving lesson last weekend. I felt that rush of ‘first day at school’ pride. But while I pat myself on the back, the technology world is moving in a different direction.

Working in tech PR, I’m often reading about the driverless car takeover. I wanted to find out whether the pounds spent learning to drive were worthwhile.

Adapting to a driverless world

As a backseat driver, I’m not a natural convert to the autonomous world. The thought of speeding up the motorway and trusting technology with my life is nerve-wracking. Last month, the government proposed changes to the Highway Code in which motorists won’t be liable for crashes and can even watch TV behind the wheel. This would be a leap out of many people’s comfort zones.

But how close are we to seeing Futurama-type motorways? The government is developing a legal framework to introduce self-driving cars safely by 2025. In some places, driverless cars already have the green light.

In Phoenix, Arizona, driverless taxis are freely available through the mobile app Waymo, which acts like an Uber for completely autonomous cars. Their fleet have had ample opportunity for machine learning — they’ve driven 20 million miles on public roads, according to VentureBeat — but there’s still a way to go before driverless cars overtake traditional vehicles.

The school for autonomous cars

One reason that fully automated cars are tricky to roll out across the board is that the real world puts drivers into complex, often chaotic situations. People can be unpredictable — we don’t always act as the robots think we will.

The Mcity Test Facility, operated by the University of Michigan, hopes to tackle this by testing automated vehicles in controlled but highly realistic situations.

It’s a mock city — a bit like a Legoland for cars — with 16 acres of roads and traffic infrastructure. Autonomous cars encounter simulated situations mimicking the ones I’ve been told to look out for during my driving lessons, such as kids playing in the street.

Research like this will be hugely important in realising safe driverless travel.

Learning human behaviour

Much of the challenge for autonomous vehicles comes down to human behaviour. While I won’t go so far as to say that human drivers are more intelligent than driverless cars, we pick up information subconsciously that current AVs might miss.

For example, AVs can differentiate between a bicycle and a car, because they are programmed to know the relative size and speed of a bike. But is this enough?

A recent study from PNAS found that AVs often neglect social cues and driver personality. Human drivers can communicate with the cyclist and read their mood — we can tell if they’re in a rush or not concentrating on the road and adapt accordingly.

Nuanced social cues will make incorporating AV into everyday life challenging.

Watch this space

We might not see a fully automated driving world overnight. However, machine learning, radar technology, smart cameras and ultrasonic sensors are improving all the time. It should be impressive to see the advancements that are made in the next few years.

For now, though, I’ll stick with my driving lessons.