Behind the scenes: finding out how self-driving cars will work
12:30 AM | September 15 2016

Behind the scenes: finding out how self-driving cars will work

Sam Mehrafzoon, Volvo Car’s Lead Engineer at Preventive Safety, explains how the sensors that make self-driving cars possible will work.

Sensors, which already play a huge role in safety in today’s cars, will be an essential component of self-driving cars. Tell us why.

 “They are the eyes of the car. We have sensors all over the cars to detect what’s going on around it. At the front there’s a forward-looking camera – which is actually made up of three cameras in an autonomous vehicle, compared with a single camera in non- and semi-autonomous vehicles. 

“Then we have a front radar and two rear radars to detect vehicles, for instance when you change lanes. We have to detect over a distance of a couple of hundred metres in order to see if there’s a car approaching quickly from behind. There’s also a Lidar – a laser scanner that detects obstacles on the road.”

Are there sensors inside the cabin?

“Yes, we also have a Driver Monitoring System (DMS) which is a camera that will detect driver behaviour – where’s the driver looking, whether they are drowsy, and whether they have stopped looking at the road.

“This technology doesn’t record anything – it’s there purely to support you. It’s my favourite because of the potential future uses. We could have face recognition as a key, for example, so you could start your car with a face scan.” 

What does testing the sensors involve?

“At Volvo we always go the extra mile, because safety is rooted in everything we do. We do tests in the north of Sweden, tests in the deserts of Nevada, as well as more extensive testing to ensure the predictability we’re looking for. We also carry out a lot of testing on the ring road in Gothenburg, which will be used for our Drive Me trial.

How do you integrate sensors into the cars?

“We have to consider how to fit the sensors to ensure that they’re in the best position to work properly without compromising the car’s design. It gets even harder with interior packaging – when fitting the driver monitoring camera, for example. How do we integrate that? We have such clean interiors with so few buttons and clean surfaces that any interruption has to be very well thought out so it’s not obvious and doesn’t make the interior look less attractive.”

So where they are placed is important?

“Yes. With exterior sensors there are issues with contamination from mud, snow, ice and so on. For example, we have to anticipate what happens when heavy snow covers the car and then turns to ice. As a result, we will package the sensors in a way to get round this without affecting the exterior design.”

What happens if a sensor fails?

“We have built-in redundancy. That means using several sensors, so that if one fails we have a backup.  It will work seamlessly without needing driver intervention. The sensors have blockage detection so they will tell the car if they can’t see and the car can ignore that sensor. It’s effectively turned off.”

Is it possible they could all need to be shut down?

“Well, the first generation of Autopilot self-driving cars won’t be autonomous in heavy snow anyway because they have to use the line markings as guidelines. In this case the system will shut down and safely guide the car to the side of the road and park. We know exactly where to stop the vehicle and where to park.”

How does the Drive Me pilot project fit into this?

“That’s the next step. Once we have verified the sensors in our own testing they will be used in the Drive Me trial starting in 2017.