Intel will be inside BMW’s iNEXT, a completely autonomous car the automaker plans to release by 2021. Intel and visual analytics company Mobileye, will provide underlying technology for the iNEXT to roam streets without human assistance.
It’s an awkward day to make this announcement, given Thursday’s news of a NHTSA investigation following the first known fatality in a self-driving car. The accident involved Tesla’s self-driving Autopilot feature, officially in beta but widely used by enthusiastic Tesla customers. Autopilot was engaged when a Model S rammed into a left-turning semi in May, killing the driver. NHTSA will take a closer look at why the Autopilot and the driver both failed to avoid the truck.
The iNEXT model is the first in a range of fully autonomous vehicles planned by BMW. BMW is betting self-driving cars will be widely used as automated taxis and in ride-sharing services. Uber already has the ride-sharing possibilities in mind and has deployed a self-driving car in Pittsburgh.
BMW, Intel, and Mobileye are also working on “highly autonomous vehicles” that may require minimum human intervention. The companies will soon test such a vehicle, though exact dates weren’t provided.
Many cars are already equipped with features like automatic braking for safe driving. High-resolution cameras, radar, lidar, and optical and thermal sensors are being added to cars and will ultimately help make self-driving cars practical.
Automakers are chasing autonomous cars as fast as tech companies like Google. Google already has a self-driving car on the street, but it isn’t perfect and has been involved in accidents. Chipmakers are providing the underlying technologies for self-driving cars.
BMW’s partner Mobileye has a proprietary chip called EyeQ5, which can make self-driving safer by grabbing, processing, and analyzing data from cameras and sensors simultaneously. Mobileye’s technology is already used in Tesla vehicles.
Intel could provide processors, FPGAs (field programmable gate arrays), connectivity chips, and security tools. Intel also sees an opportunity to sell more server chips as the demand for deep-learning systems grows. Self-driving cars are expected to rely on deep-learning systems in the cloud to teach vehicles to identify objects or signs.
“Cars—and everything they connect to—will need super-powerful, secure and reliable electronic brains that make them smart enough to act like human drivers,” Brian Krzanich, CEO of Intel, said in a blog entry.
Intel this year acquired a string of companies, including Yogitech and Itseez, that bring computer vision and safety features to cars.
BMW competitor Volvo is building an autonomous car with Nvidia, which offers chips and computers that can detect signals, objects, pedestrians, lanes, and signs. Nvidia provides a powerful water-cooled computer called Drive PX2 that is stored in the trunk of a car.
There are still legal and standards questions about self-driving cars. Mapping firms and the auto industry are forming standards around exchanging sensor data such as road conditions via the cloud.
About 21 million self-driving cars will be on the road by 2035, predicted research firm IHS.