Search This Blog

Tuesday, March 18, 2014

Volvo Designs Magnetic Roads For Cheaper, Simpler Self-Driving Cars

As reported by Extreme Tech: There are myriad visions for a future filled with self-driving cars. For example, there’s Google’s experimental driverless car bristling with sensors, as well as more modest systems that would only take over from drivers for short periods. The problem with more ambitious self-driving car technologies is the considerably higher cost, whether in public infrastructure (networked roads) or the smarts built into the vehicles themselves. Volvo thinks it has an idea that could make self-driving cars work with much less hassle. All we need is a bunch of magnets embedded in the road.

Volvo began developing its magnet-based smart car system after looking long and hard at the other proposals on the table. It’s not just the cost of advanced sensors, cameras, GPS, and LIDAR that make self-driving cars tricky, the reliability is also questionable. Electronic solutions are more prone to failure in general, but even more so when inclement weather strikes. A magnet? Well, that’s always a magnet, and it can be paired with other automated technologies to make a fully driverless car.
In order to test the idea of using magnetic roadways, Volvo actually built a 100-meter test track in Hällered, Sweden and raced a specially modified S60 down it at over 90 mph. Engineers lined the road with neodymium magnets (20mm x 10mm) and ferrite magnets (30mm x 5mm) in lines down the edges and middle of the lane. The company tested both embedded and surface installation, finding that magnets on the surface would be effective and easier to install. Although, either option is sure to cause headaches in the case of roadwork.
Magnetic sensors are nothing new, but at the speeds we busy humans often need to drive, existing hardware wasn’t sensitive or fast enough. Volvo engineers calculated a car would require at least 400 magnetic samples per second to remain on the straight and narrow — a regular magnetic sensor can only do about three readings per second, and even then only when it is within a few centimeters of the magnet. So Volvo decided to roll its own magnetic sensor rig with five sensor modules, each with 15 smaller Honeywell magnetic sensor pods. This rig was attached to the bottom of the car and was able to pull in 500 readings per second.
Volvo SenorsThe system was able to monitor the car’s location to within 10 cm at 45 mph when telemetry factors such as speed and acceleration were figured in. You’d probably want the precision to be a little higher before taking your hands off the wheel, but you get a lot for your money here. The advanced sensor package on Google’s self-driving car has about $150,000 worth of sensors, but Volvo estimates its magnetic sensor package will add only $109 to the cost of a car when produced in large quantities. Volvo also claims installing magnets in typical two-lane roadways would cost an average of $24,405 per kilometer. If that sounds like a lot, it’s not actually bad in the context of self-driving technology. Of course, you could only use this system where the magnets had been laid down — Google’s car works almost anywhere right now.
As the technology for self-driving cars becomes a reality, we need to ask ourselves how smart the cars should be. Expensive sensor packages are great for completely controlling a vehicle so you can take a nap, but only in good conditions. A bit of ice or some fog could make things awfully sketchy. If we rely on magnets in the road (or some other passive tech) everything is more reliable, but possibly not as convenient. A networked on-board system can respond to traffic dynamically and provide detailed analytics. Magnets — they just keep you on the road. However, it might end up being more important to focus on what’s feasible than what’s clever in the end. Magnets could end up as part of a more advanced system that at least has a basic fallback mode when things go wrong.