With the vision of a fully autonomous car coming into clearer view, the semiconductors and sensors used in these innovative vehicles are emerging, improving, and becoming more affordable. From LIDAR to image sensors to solid state image scanners, the technology that goes into a fully autonomous car is improving every year and helping to make the day when your kids can hop in a driverless car to get to soccer practice a reality.
So far this year there have been several major developments on the autonomous vehicle front. At the North American International Auto Show in January, Transportation Secretary Anthony Foxx announced that the government was removing hurdles to developing such vehicles and setting further guidelines for their use. Around the same time, the Obama administration proposed spending $4 billion on self-driving cars. If approved, the proposal says the money will be spent over a 10-year period to “finance research projects and infrastructure improvements tied to driverless cars,” according to The New York Times.
Stepping up to the Plate
Carmakers also are making important inroads with their autonomous vehicle efforts. Jaguar Land Rover, for example, kicked off its MOVE-UK project to support the development of safer and more effective autonomous vehicles. The company’s employees are driving around the streets of London every day as vehicle sensors record how they behave and make decisions in various real-world driving scenarios (including heavy traffic, busy junctions, road works, and bad weather).
In another key development, U.S. vehicle safety regulators announced in February that the artificial intelligence piloting Google’s self-driving car could be considered the “driver” under federal law, according to a Reuters report.
“Since the National Highway Transportation Safety Authority (NHTSA) is willing to call the car a driver, it seems to suggest the current safety standards could be tweaked to allow for the production of fully autonomous cars, where a person doesn't even have the option to grab hold of a steering wheel or slam on the brakes,” the report says.
Driving Down the Costs
With both industry and government embracing and supporting the vision of autonomous vehicles dotting the nation’s roadways, the semiconductors, sensors, and other components used to develop these cars are becoming increasingly sophisticated, available, and – in many cases – affordable. The rotating mechanical LIDAR scanners bolted to the top of the original Google cars, for example, cost $50,000+ each and were used for development (not production) purposes. This is changing.
Today, this important element of the autonomous car is much more affordable, according to Akhilesh Kona IHS’ analyst, automotive semiconductors. At the Consumer Electronics Show (CES) in January, Quanergy introduced its new S3 solid-state LIDAR system designed to bring “versatile, comprehensive, and affordable sensing to autonomous cars.” The S3 is small, has no moving parts, and will cost $250 or less once it goes into production.
Kona says he sees Quanergy’s new technology as a major breakthrough for autonomous vehicle development. Up until now, Kona says autonomous vehicles relied on mechanical-based components and laser beams that were “thrown onto” a scanning meter. That meter, in turn, would throw the beams out to collect various pieces of data.
“Because LIDAR was so costly, Google, Ford, and others could only invest in it for development,” Kona explains. “At this new, low price, we should start seeing the LIDAR used on production models and in volume.”
What’s more, “the new technology generates data faster than the bulky modules currently mounted on cars,” says Kona, and provides the diversification needed to discern among different physical principles (for example, a curb versus a pedestrian versus a cyclist).
“Radar can determine that distance to an object and a camera will tell you what the object is,” says Kona, “but in some cases you need confirmation of this information from another type of sensor. That’s where LIDAR comes in.”
Originally working with Mercedes-Benz to develop, test, and deploy its LIDAR-based systems, Quanergy has since formed partnerships with Hyundai and other automotive OEMs, according to Kona.
A Mixed Bag Ahead
For the year ahead, Kona says components buyers should keep an eye on camera module pricing – an area that could increase as the related technology improves.
“In general, semiconductor prices decrease year-over-year as the technology improves and as the manufacturing process becomes more efficient,” he explains. “However, when it comes to camera modules for autonomous cars, the prices on the processors used in those applications are actually increasing.”
Kona points to Mobileye’s vision-based, advanced driver-assistance systems as a good example of this.
“The average selling price of its processor is increasing year-over-year,” says Kona, “as the number of features and functions that it has to handle continues to expand.”
Conversely, he says radar sensors are dropping in price as manufacturers such as Fujitsu and Panasonic move away from using costly silicon-germanium technology.
“We could see a 40 to 50 percent drop in transceiver costs,” says Kona. “This, in turn, will allow carmakers to incorporate more radar into their platforms.”