Why we’re still years away from having self-driving cars

Autonomous vehicles have been hyped for the last few years, and while great advances have been made, we are still a long way from the widespread adoption of self-driving cars.

Of course, there have been huge advances in technology to help drivers on large roads like motorways and highways. These advances include adaptive cruise control, auto emergency braking (which my car often does for no apparent reason!), lane assist technology, and rudimentary autopilot systems. 

This technology works so long as the driver is still paying attention, and nothing out of the ordinary happens on the road.

We were sold the dream of cars that could talk to each other, so that the car could take you safely to your destination while you relax in the back seat. The promise of self-driving cars would change the car ownership model, and you would need fewer cars per household as the car could be shared with others instead of sitting on your driveway all day. Your car could take you to work in the morning, then drive home and take another person in your family or on your street to their work. Self-driving cars would result in fewer cars and perhaps less ownership - leading to more rental of car capacity.

Flash forward to today, and precious little has changed about our daily driving. You probably hear a lot less about self-driving cars than you did a few years ago, and the prospect of safely dozing off behind the wheel on long drives remains a distant fantasy, even if old-school carmakers are working with startups like Waymo, Cruise, Argo, and Zoox on the technology.

However, we are still a long way from ubiquitous autonomous vehicles. While main roads have been well-mapped and are fairly predictable, there are a lot of edge cases that machines do not know how to deal with. The well-known debate of asking a computer to choose if they had to crash into an elderly person or a child is one such moral dilemma. Also, humans drive in fairly well-understood patterns, but sometimes they are haphazard and break the rules. This can be hard for a computer to deal with in real time -- until there is a huge volume of use cases from which to draw.

The dream of automated driving remains a very real proposition, but we have to be realistic in our time frames. There is good agreement from analysts that when 100 percent of the vehicles on the road are fully autonomous, then it will be much easier for computers to deal with edge cases. The issue is that this will take a long time, and there will be an extended period where some cars are autonomous and some are driven by humans. 

I believe that as computing power improves, artificial intelligence improves, and we get better at creating learning computers, the dream can be achieved. Sensors, tracking, scanning and monitoring equipment in cars become more compact and affordable each year. These sensors are starting to generate huge amounts of data from which car manufacturers and software developers can learn.

I truly think that we will see autonomous cars in my lifetime, but I think my kids will need to learn to drive and pass their driving test the old-fashioned way. Human power is still going to be around for a while.

The link below was the inspiration for this article: