ARMv8-R Architecture for Next Generation Automotive

This is the third of three blogs I've written around the announcement of the new ARMv8-R architecture for real-time embedded processors.  Simon Craske, our Embedded Processor Architect, and myself will be presenting more details at the ARM TechCon conference in Santa Clara next week...

To my mind, automotive electronics is at an exciting time with modern cars providing increasingly useful features and with the promise of more to come.  Electronics, and therefore embedded systems, are providing increased performance, efficiency and functionality for engine management, powertrain, chassis control, body electronics and In-Vehicle Infotainment (IVI), and these capabilities have become the norm in the modern vehicle. After all, who would buy a car now that didn’t have a nice entertainment system with a media player connection, good performance and fuel efficiency and, above all, a car that was seen as comfortable and safe for drivers, passengers and those outside.

However, with these features comes additional cost and complexity:  The average car now has around 70 Electronic Control units (ECUs), a Car Area Network (CAN) bus connecting them and inside there are processors running many megabytes of code in total.  Electronics accounts for a double-digit percentage of vehicle unit and development costs and this is rapidly increasing, together with cost of ownership for servicing, software upgrades and, dare I say it, bug fixes.  It’s all a far cry from the cars we drove ten or twenty years ago where there was just one electronic control unit for engine ignition.

Much of this is driven by purchasers’ demanding increased comfort and convenience but there is also a background of legislation designed to make vehicles more energy efficient, cleaner and safer, with the overall goal of reducing accidents and injuries. Examples of this include vehicle stability (anti-skid) systems and automated pre-crash braking and air bag deployment.

As this trend continues, automotive manufacturers and their Tier-1 suppliers are having to deal with increasingly complex and sophisticated software and system development which often involves model-based automated code generation and integration of software from different sources in the supply chain.

And as far as I can see, this trend is accelerating with the introduction of Advanced Driver Assistance Systems (ADAS) such as lane departure warning, blind spot detection and dynamic cruise control.  I have these on my car now and I must say they're features I've quickly come to like.  The cruise control in particular impresses me; it senses vehicles around you, your road positioning using radar and video and it works out what you’re doing based on acceleration, braking, steering angle, use of indicators etc.  And, boy does it work well!  On Monday, I set it to 70 miles an hour on the A1 and drove for over two hours without having to touch the accelerator or brake at all; even when the traffic slowed down to a crawl at one point.  Overtaking wasn’t a problem for it, and it was braking and accelerating just as well as I could, if not better.  Of course you have to keep foot available to take over if needed, but I didn’t need to once.

To me though, this is just first generation ADAS.  It must have had a pretty powerful processor and image recognition system working quite hard, and of course it has to be a safe system, probably engineered to a high Safety Integrity level with error management in case of an electronics or software fault.  As I see it, the next generation of ADAS will also be communicating with other vehicles and the roadside by wireless and these systems are in development now; they call them V2X meaning Vehicle-to-Vehicle and Vehicle-to-Infrastructure Communications.

V2X is being trialled now, in the USA and in Europe and probably in Japan for all I know.  You can read about it on the US Department of Transport web site and other places including the Car2Car consortium and on a number of company web sites such as Autotalks, who are building the required semiconductor devices.  The benefits of V2X are pretty obvious once you think about it; transponders on each vehicle and around the highway can warn of unseen congestion or danger ahead, provide braking assistance such as when approaching a red light and give a warning if you might pull out in front of another vehicle and so on.

The potential of ADAS and V2X to save lives and cost to their economies isn’t lost on politicians and both the European and US governments have set vehicle manufacturers the objective of there being zero collisions by the year 2030 or so. Personally, I think that with the technology we have on offer this is doable but we have to get on with it as these system will take a long while to develop and deploy in production vehicles and along all major highways.

Of course these technologies will increase the amount of processing required in every vehicle and at the roadside which, as vehicle production approaches 100 million a year and with each vehicle having 70 or so processors, indicates a substantial market size.  Various ARM processors suite the different applications with high end Cortex-A50 series of application processors in IVI and at the front end of the ADAS, Cortex^®-R processors in chassis, powertrain, engine management etc. and Cortex-M processors in sensors and body electronics such as window lifters and side mirror adjusters.

The new ARMv8-R architecture for real-time Cortex-R processors will play an important role in this automotive future.  ARMv8-R will provide vehicle manufacturers and their Tier-1 suppliers with a processing platform for these kinds of applications where complex software can be consolidated safely as described in my previous writings on the topic.  These systems must demonstrate Functional Safety, i.e. an ability to continue operating in the presence of faults, because drivers, passengers and other road users will come to rely on them.  For sure, the ARMv8-R architecture will contribute towards meeting these requirements.

I started off saying this was exciting, and one reason for that is these systems do not necessarily imply driverless cars or similar.  If anything, I think driving can become more enjoyable as it gets easier, safer and less frustrating thanks to the reduction in highway congestion and delays caused by accidents. Drive safely please.

CBT

Embedded