NXP's contribution to the self-driving future:
NXP is a subsidiary of Philips, the number one Dutch technology vendor. After its purchase of Freescale Semiconductor in 2015, NXP became the leading semiconductor manufacturer in the world, counting more than 45,000 employees, shipping about 30 million processors for automated driving-systems to eight out of 10 of the world's top vehicle manufacturers.
Although NXP specialises in making microprocessors (MCUs) for cars (15% of all in-car processors are NXP-made) it also develops V2V/V2E technology, IoT, connected car (security) applications and even launched a full-blown computer for autonomous driving, the BlueBox. Will it ever come to a NXP self-driving car too?
Advanced Driver Assistance Systems
Every impulse has to be processed before it can be understood. In a car this means every button pushed, every drop of gas used and every inch of road traveled. For a driverless car this means every surrounding object within a 500 foot radius. NXP builds processors for almost everything, from car radios and automatic windows to gas tank analyzers or tire pressure monitors, but for driverless cars in specific, contributions go from sensor processors to video transmitters and network transceiver. These are some examples of NXPs most commonly used microprocessors for driverless car technology:
The S32V230: used for image processing and sensor vision fusion. It is the first automotive vision system-on-chip (SoC) with the reliability, safety and security measures to automate and a self-aware car. In Advanced Driver Assistance Systems it can be used for pedestrian detection, object detection, lane departure warning and traffic sign recognition.
The MPC577xK: delivers a solution that helps to move radar into mainstream vehicles and other satellite radar nodes, in addition to the forward facing radar node. Can be used in combination with the MR2001, NXPs 77GHx radar transceiver.
Connected car security
And lastly the MPC560xE: Today's cars can have 200 or more networked components. This particular one is designed to move camera-based data via Ethernet to a receiving system and vice versa. This MPU is used in NXPs Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) solutions. NXP is a leader in v2E technology through its many CAN, LIN and FlexRay products. And projects. Together with Cohda Wireless, NXP started RoadLink, a plan to own all future V2X trails. Another noteworthy IoT project NXP joined with the US Department of Transportation was the Smart City Challenge.
NXP might be the leading semiconductor manufacturer in world, there are other competitors on the horizon. Nvidia, mainly known for its computer graphics cards, has also been spotted delivering graphics processors to the automotive industry, for example the Tegra X1 used by Audi and Volvo or the Tegra K1, implemented in the famous Tesla Model S. Another watchful privateer is Intel, global leader in computer processors, who teamed up with BMW and Mobileye in 2016, looking to create a fully autonomous NXP self-driving car together by 2021.
If NXP ultimately loses the sensor/processing/V2E war, it might be able to count on another branch. NXP is investing heavily into smart car security. From NFC protection and smart cards to USB type-C, network protocols and RFIDs. The first fully autonomous cars will be an attractive target for hackers, but NXP will give them a run for their money.
NXP (together with Cisco Systems) eventually put all those separate ADAS processors into one central computer, the BlueBox. It is the ultimate impulse fusion processor, tying together all sensors found in an autonomous car including sensors, cameras, Lidar radar and even V2X systems. Mind you, the sensors and cameras are excluded, the BlueBox is ‘only’ a processor.
But NXP claims the BlueBox can be implemented by all car manufacturers in all car systems, using whichever sensors and camera they want. On top of that is the software Linux based an open source. It i ready for OEMs and Tier Ones to develop and experiment their own autonomous cars", says Matt Johnson, NXP’s vice president and general manager for automotive microcontrollers and processors. “And the platform can be taken straight to production.”
The BlueBox can process 90 million instructions per second at less than 40 watts. This means the system uses less power than biggest competitor Drive PX2, developed by Nvidia’s, which consumes nearly 250 watts. NXP claims 4 out of 5 major automakers implemented the box straight after launch and thus the BlueBox goes head to head with the entire market; companies like for example Delphi, who you remember building Audi’s zFas processor. But while the BlueBox was welcomed with enthusiasm critics immediately mentioned that it was built disregarding mapping, big data and deep learning, which according to them are needed to build a truly driverless car.