Managing automotive system complexity is further complicated by software, as well as tracking requirements.
Just as the overall system complexity is causing ripples through the automotive supply chain so too is managing the system complexity, with software in particular.
With so much new technology, and so many new ideas to keep track of, it would seem a huge undertaking by the automotive OEMs. In the midst of making decisions about the usual incremental improvements, the system architecture decisions about how to move forward to make vehicles truly autonomous are being worked out now.
Making things even more complex, Steve Roddy, senior group director for Tensilica marketing at Cadence reminded that the traditional automotive OEMs have never been known as aggressive risk-taking, fast-moving companies. “It’s not in the nature of a 100-year-old car company but you see some of these bigger guys that are doing both. The smarter ones have realized, ‘Oh crap I don’t know what the right answer is,’ so you’ll see incremental deployment of technology in current 2016, 2017, 2018 model cars. You’ll see some that are now shipping the 802.11p automotive Wi-Fi peer to peer networking stuff in some of their brands which we happen to know because Tensilica is being used in the NXP chip and they are the first on the market to do that.”
Likewise, one of the big Detroit-based OEMs is not normally thought of as being cutting-edge, so it’s common to see incremental deployments. However, this particular OEM also hedged its bets and bought a San Francisco-based startup working on fully autonomous technology.
To be sure, automotive OEMs can’t afford to not make changes.
Glenn Perry, vice president and general manager of the Embedded Systems Division at Mentor Graphics and the head of Mentor Automotive pointed out that going back probably four years ago, some of the most advanced OEMs who were well aware of their limitations in terms of software and electronic design started to — instead of writing their usual paper spec to shop around with various Tier-1s — build hardware and software prototypes which were pretty close to what was intended for production. So instead of a paper spec, they’d use the prototype as essentially the RFQ to go out to the Tier-1s.
Further, he noted that there was a time not long ago where the notion of an OEM startup was just unheard of, and what changed is the OEMs’ one technical area that was a big barrier to competition of new entrants: the combustion engine technology. “However, if you’ve ever seen a Tesla with the body off of it, you can appreciate the simplicity of an electric vehicle in comparison to a combustion engine. Therefore, you have dozens of new OEM electric vehicle startups that are also competing.”
This is changing the game fundamentally.
Back down into the nitty gritty, with higher levels of complexity, making the right tradeoffs is a delicate balance.
Roddy observed: “You’re trading off system complexity and we haven’t even touched on software complexity. If you put all the software into one big box on the dashboard like a tablet, or you stuff a whole bunch of CPUs or GPUs behind that, you’ve got lots and lots and lots of software. How do you get that certified for functional safety grades, and avoid all the lawyers down the line if it’s 5 million lines of code? Can you ever really prove that you did it right or is some jury in West Texas going to slap a big fine on you?”
Particularly because everyone is overly familiar with the rate at which PCs and phones need to be rebooted, some in the automotive ecosystem are driven by looking at that particular aspect, he continued. “I’ve talked to some carmakers or system makers who are terrified of the idea that they have to be responsible for doing all of that validation, and proving out that they’ve done all of the ASIL D- level qualification to prove there are no software bugs. On the other hand some others turn around and say, ‘At least if I’m doing it, I control it.’ I have no idea even if my vendor says they are certified, do they really follow all the same practices? Do they really all follow the same coding standards? What a pain if I have to dictate the coding standards, and I have to have auditors go in and inspect the code especially since I’m not willing to buy all the code but I trust that they do it and they ship me a firmware module well then I have to have auditors go in and inspect that just like if I had to do it myself — maybe I’d be better off doing it all myself — writing it all myself — integrating it all myself —if I have to test it.”
Today, there’s so much uncertainty here, compounded by dozens of new players because there’s so much new. “Everyone realizes it’s another $80 billion worth of electrical components, chips, panels, wiring that’s all going to be shipping in cars 10 years from now compared to today and everyone wants their slice of it so,” Roddy added.
One thing is for sure in all of this, there are still many technical, legal, societal, and other issues to be worked out. How it shakes out will be revealed over the next five years. It’s sure to be an exciting ride.
Cars, Security, And HW-SW Co-Design (part 2)
Standards are helping address some issues in concurrently designing hardware and software. More challenges are ahead as automotive electronics and cybersecurity issues enter into the equation.
Lawyers, Insurance And Self-Driving Cars
Semiconductor companies have to contend with more than technology when it comes to the automotive market.
Self-Driving Cars Rattle Supply Chain
Shift from hundreds of MCUs to more centralized computing creating chaos in auto industry.