Daydreaming of the autonomous car

Daydreaming of the autonomous car

On a recent drive to visit relatives in Bologna Italy, I found myself longing for the autonomous vehicle.

My journey from Munich through the Alps is full of spectacular views and breathtaking snowcapped mountains, which unfortunately, I often must forgo as the driver. Cruise control is a helpful tool in managing my speed, but the driver’s mind must still be fully focused on driving. I often find myself fantasizing that my car will take the wheel – monitoring the distance between cars, adjusting speed, and steering automatically – while I take in the surroundings that, until now, I’ve been forced to overlook.

As I drive, my mind continues to wander: Are we truly ready for a fully autonomous vehicle? Technically? Legally? Architecturally? Psychologically?

As a engineer, and an automotive enthusiast, thoughts such as these cross my mind on a daily basis. A car with the ability to drive people of all ages, conditions, and needs would be a truly radical social change. People could ‘work’ while commuting (or simply enjoy the landscape), families could choose to live further out from the cities in greener, healthier, and less dense areas. Physically impaired drivers could maintain greater mobility, limiting the cost on taxpayers while opening up new freedoms. Pollution could be reduced from more efficient driving methods on major roadways. As we are looking at the autonomous sphere, what gaps must we fill? And what is the role we have in all of this?

The road environment is densely populated with actual or potential hazards — other vehicles, people, weather, changing situations — and autonomous cars must be in constant communication with each other in order to keep the driver safe. From a communication/infrastructure perspective, which communication link and protocol will emerge to allow this vehicle-to-vehicle or vehicle-to-infrastructure system to become a reality?

  • Dedicated Short Range Communication (DSRC) was conceived for auto in the form of 802.11p, which can theoretically handle up to 2,000 connected vehicles with a total bandwidth. Although contained and manageable, the absence of a standardized infrastructure has left the technical community skeptical of this technology.
  • LTE and even more powerful 5G technologies bring the promise of capturing a large amount of data with shorter, more controlled latency. The technical community has raised questions whether 5G providers will allow part of their highly profitable bandwidth to be used to ensure safe operation, essentially trading safety for profit.
  • Even the common GPS signal can be challenging. GPS signals must secure a resolution of one meter or less, and the algorithms must use the statutory quality of the signal to be able to infer decisions before having a legitimate safety aspect.

Developing an electronic control unit (ECU) for autonomous vehicles implies using state of the art standards, like ISO26262, but product liability and legal relevance remain the “holy grail.” We need more platforms that can enable multiple partners to develop products and innovative ideas behind the initial concept. It is not only the hardware but the entire ecosystem that counts (i.MX line is a great example of this).

It’s possible that in ten years’ time, I will be enjoying a panoramic view of the Alps while traveling to my homeland? At least I hope so. In the end of course, it’s about how we make the most out of our available time with our family and make our lives safer and more comfortable.

Our approach to the autonomous vehicle

To build the self-driving car will be a collaborative effort. No one company will reach this goal on its own. We’ve selected a broad range of strategic partners, such as Green Hills Software for its strong competence in secure and safe operating systems, and Neusoft for its competence in object cognition software technology.

In the autonomous vehicle, the capacity to process quickly and efficiently is a key element to success. We have developed fundamental processing blocks to accomplish this with our radar signal processing toolbox and image signal processing block. We also license the APEX engine for image cognition processing from CogniVue, for which we are creating an entire software development kit.  UPDATE SEPT 2015: We acquired CogniVue and its APEX engine for image processing. We are creating an entire software development kit.

Also, leveraging the latest GPU from Vivante ensures diversity of hardware and enhanced graphics capabilities.

The decision to support our heterogeneous integrated IP set with a common standard language, namely OpenCL adopted by APEX, GPU and ARM neon, should give a sense of vision – openly enabling any potential software developer by using a nonproprietary language dedicated to parallel processing.

Lastly, Layerscape family allows us to offer the entire range of computational depth where needed without diminishing our strong safety proposition; and is competent in both network connectivity and system high bandwidth inter-processor links. We complement this with 24-7 reliable networking technology.

The road ahead in perfecting ADAS technology for the fully autonomous car is long and winding, yet we are well equipped to compete, thrive and succeed.


At NXP, innovation is always now, but our focus is always the future. Our dedicated team of experts is united by a passion to make everyday life more remarkable through technologies that continually redefine life as we know it.

Comments are closed.

Buy now