Autonomous cars are developing rapidly, and it looks like they’re here to stay. With over a million deaths each year attributed to road accidents, it’s fair to say the increased safety that self-driving cars promise to bring is much needed. But whether we’re ready’s another matter; so here we take a look at five problems with autonomy and how we might be able to solve them.
1. Predicting the unpredictable
Computers generally work like they’re supposed to, which applies to the artificial intelligence that’s going to pilot our self-driving cars. However, when humans are at the wheel making rash decisions, unpredictable things often happen. For now, people are still better at reacting to emergency situations than computers, but the answer to this problem could lie within vehicle-to-vehicle communication.
V2V will link cars up to each other, allowing them to communicate their position, speed and direction, but can only be truly effective once most of the cars on our roads are equipped with it.
2. Making the right call
In the event of something unpredictable happening, driverless cars have to make a call. Let’s say a child suddenly runs into the road, and the only way for your car to avoid them is by swerving off the road straight into a tree. Your car’s got to decide whether to hit the child or potentially risk your life instead, and if the wrong call is made there’ll be serious repercussions and questions.
Ethical issues like this are key to how engineers and technological whizzes are building autonomous tech, but if they can figure out algorithms where your car makes a predetermined choice that’s always going to be the right one, then road safety will be massively improved.
3. Red tape
With problems like this hanging over self-driving tech, it’s little wonder that there’s a lot of legislation that needs to be passed before autonomous cars can take over. Different countries are going to have to work on their own laws, and our cars might have to adapt to this too.
Apart from our lawmakers and bureaucrats dealing with the red tape efficiently and decisively, the technology itself is going to be the most important factor in overcoming this hurdle. Better, more reliable systems and sensors will help to prove that driverless cars can work safely, allowing them to be freely used on the roads.
4. Mapping it all out
Whether we’re talking about just the UK or the whole world, there’s a whole lot of roads out there. Fully autonomous testing currently revolves around driving on roads that are mapped out to the inch, as well as the array of sensors fitted to the cars. Hardly any roads have been mapped out to this degree, barely scratching the surface of what needs to be done. Not only this, but roadworks and layout changes could render what’s already been mapped as outdated.
Companies such as Here are currently doing all they can to map as many roads as possible, but the best way to make sure that self-driving cars are safe even if a map’s wrong, is by perfecting how they read everything that’s around them and coming at them.
5. Seeing where they’re going
And here lies one of the biggest challenges facing autonomous cars, because at the moment the tech being used often struggles to read situations perfectly. From going over a bridge where the lack of surrounding buildings and landmarks causes problems, to failing to recognise whether a dark patch on the road’s a puddle or a pot hole, there are lots of issues to be ironed out. The cameras used to keep you in the right lane can struggle when road lines are worn out or not present at all, while adverse weather and even shadows can cause problems too.
The answer to questions like these comes solely down to the sensors and tech being developed in driverless cars. As well as improvements to features like light sensing radar systems, or lidar, smart roads might be able to help, warning your car about what’s ahead and potential hazards like traffic, accidents and even poor road surfaces.