You can’t see the future by looking sideways
Richard Sun, General Manager, Digital Cockpit Product Line, ECARX
Richard Sun, General Manager, Digital Cockpit Product Line, ECARX
OEMs can’t expect to stay ahead if they look at software and hardware in isolation. Ever-increasing levels of complexity within the digital cockpit are compounded by the global shortage of software engineers, making it virtually impossible for them to do everything in-house: leveraging the full capabilities of their key suppliers is essential.
I began my career in 2002 in the R&D department of a mobile phone company, before moving to a speech recognition and AI company. During that time, I built up a lot of knowledge and expertise with all aspects of human machine interface (HMI), product definition, solutions architectures and mass-scale projects based on IoT. I continued to develop in those areas when I came to the automotive industry ten years ago. Since then, I’ve been amazed at the transformation that’s taken place within infotainment, and how quickly it’s still changing.
Looking back, some interesting milestones stand out. Apple launched the first iPhone in January 2007: it wasn’t the first to use a touchscreen but did so in a way which transformed consumers’ acceptance of them as the best HMI for mobile devices – buttons were on their way out. In the summer of 2011, an influential article was published in the Wall Street Journal, titled ‘Why software is eating the world’ and it described how silicon valley tech companies would transform the way we do things, how it would change industries that function primarily in the physical world, and correctly predicted that ‘the trend toward hybrid and electric vehicles will only accelerate the software shift.’ Later that year, Toyota revealed the Fun-Vii concept, defined as ‘A smartphone on wheels’. Whether or not you agree with the sentiment behind that, it’s certainly become more closely aligned to the way in which many consumers view vehicles now. In 2014 both Apple CarPlay and Android Auto were announced. By then, smartphones were definitely conquering the world.
Root cause analysis
Yet despite these events, infotainment really wasn’t a priority for OEMs. Systems were just part of the interior design and even then, not a major differentiator within it. You’d have the audio system and perhaps a touchscreen, but they were far from common and even then, they weren’t always standard equipment – you’d have to pay extra. For all of its many other qualities and attributes, the vehicle was sold as a means of transport, basically moving people from a to b.
Because infotainment wasn’t seen as important then, the way that systems were developed was basically for the OEM to write a specification – nothing very fancy – and give it to a traditional Tier 1, who would then be entrusted to deliver against it. The OEM wasn’t interested in the Tier 2 or the rest of the supply chain: they just wanted the box, which they’d validate, and that was pretty much it.
Up until 5-7 years ago, OEMs were still trying to do things this way, even though we’d already moved to more complex systems, and the box was getting bigger and bigger. They were still thinking the traditional way. The real issue was that for the Tier 1, their business model was selling hardware, and because their customer was the OEM and not the end-user – the person actually buying the vehicle – they weren’t really interested in the requirements but just fulfilling the specification.
Now, of course, it’s very different. Even entry-level vehicles have sophisticated infotainment systems, and digital cockpits are one of the defining elements of any vehicle interior, especially EVs. And consumers always want more: they increasingly view vehicles not as mere transport, but more as entertainment machines – particularly in markets such as China, where consumer expectations in terms of the amount of technology within the vehicle can be very different to Europe or North America.
This has driven radical change in how OEMs develop infotainment and digital cockpit systems, and the way in which they leverage their supplier base in order to deliver the innovations their customers expect, while meeting ever-more demanding cost and time-to-market pressures.
Putting the customer first
A very interesting and important part of the discussion is that, traditionally, Tier Ones were never really focused on the user experience, simply because their business model lay solely with the OEMs and not with the end-user. That’s all changed now, and OEMs place far greater emphasis on what their customers want. This change in strategy has seen OEMs – with increasing speed – taking back control of the whole cockpit user experience.
This is especially true of EV start-ups, who see this as their number one priority, perhaps because they’re the most cognizant of the impact the Tesla Model S had when it was first launched and have taken that learning onboard far more readily than some established manufacturers.
In other words, everything now is much more end-user-focused, and that has made OEMs change their minds about how they leverage suppliers – that is to say, there’s a realisation and acceptance that they can’t just give a specification to a Tier 1: this doesn’t give their customers what they want. Which is why every OEM now spends a lot more time and investment not only on their Tier 1s, but also with the Tier 2s and even Tier 3s, to ensure that the system will deliver exactly what’s required.
At the same time, they’re also adding more in-house resource to take greater responsibility for the areas that are becoming increasingly important to their brand DNA: the HMI itself is probably the most significant of these, with ADAS being another, because they know how important they are to their customers, and how strongly they can differentiate themselves from their competitors, creating a safer and more enjoyable environment.
This means, primarily, increasing the number of software engineers they have, but the question is: where are they going to come from? Already there’s a worldwide shortage of them within our industry – especially ones with the right skill sets and experience – and it will take time for that gap to be filled by new waves of graduates. Then there’s the issue of attracting those engineers when there’s so much demand elsewhere.
In many cases, OEMs have sought to recruit software engineers made redundant by the big tech companies such as Meta and Twitter, but it’s just as likely that they’ll be drawn to start-ups and other tech companies in preference. Lastly, there’s the sheer complexity of vehicle software, which is not always fully appreciated by specialists from other industries.
Even if you do think ‘I can recruit enough software engineers’ you must ask if this can ever be cost effective. Take mobile phone companies as an example: they’re the ones who really afford to have big in-house development teams because they can spread the cost over a disproportionately high number of units. To illustrate that, Toyota sold 10.5 million units in 2022, making it the world’s largest carmaker – in the same year, Apple sold more than 220 million iPhones.
There’s also a cultural element: even if the stereotype of the latte-sipping, shorts and Birkenstock-wearing software engineer is not entirely true, there is still a different cultural mindset from the traditional automotive engineer, who happens to do software. Products are never entirely finished but continuously updated and improved.
We must also take lessons from OEMs who have set-up in-house software divisions. The idea is good, but the results achieved have been mixed, with some doing very well, while others have struggled, with software being a major contributor to vehicle programme timing slipping significantly.
In almost every case, no OEM or supplier can do an end-to-end software solution for a new vehicle: they’ll always need to work with a partner because it’s so complex, with expertise required not just in the HMI itself but Autosar, automotive-grade Linux, QNX, Android, hypervisor, middleware and the ecosystem. Try to do this alone, while also constantly looking around at what your competitors are doing, and you’ll lose your speed and your efficiency. OEMs realise this, but at the same time, they know that they can't go back to the traditional Tier 1 model – it just won’t work any more.
Finding the right balance for the future
This is why as a relatively new supplier, we’ve structured ourselves from the outset to offer the complete range of capabilities to our OEM customers: hardware, software, OS and even SoC – all in-house. We also work seamlessly with other leading suppliers and can function as a Tier 2 or a Tier 0.5 to enable fully flexible product development of world-class digital cockpits.
One of our most recent innovations, coming to market in late 2023, is the ECARX Antora 1000 – our fourth-generation digital cockpit computing platform. It features the 7nm high-performance SoC we designed in collaboration with SiEngine, running on our own operating system and software stack. The highly optimised design also makes it simpler, more compact, and more energy-efficient than its predecessor, while also enabling OEMs to reduce development times by 20 per cent. The platform is modular, delivering the degree of standardisation needed to be cost-effective, while also enabling the level of customisation necessary for OEMs to realise the look and feel which defines their HMI and branding. It’s difficult to achieve this balance, but we’ve worked very hard to reach it.
We’ve also invested heavily in the underlying software stack to optimise the full capability of the hardware. One example is ECARX Antora 1000’s HiFi DSP: very strong, very powerful – we don’t need an external DSP at all. So, this saves the OEM cost and packaging space for the head unit, but we still make money from the software optimisation – a win-win situation.
The approach with our fifth-generation digital cockpit platform, Makalu, is quite different: we really wanted to reach our summit so far with this technology. This makes it more expensive, as we integrated the highest possible computing power. It’s not uncommon for OEMs to have to upgrade hardware every couple of years in order for the system to have the performance required for all of the planned software over the air (SOTA) and firmware over the air (FOTA) updates. This is costly and inefficient.
So if you only want to change the head unit every 5-7 years you need a really powerful SoC to ensure you can provide a consistent user experience that matches your brand DNA while at the same time offering more and more features over time, allowing you to really leverage OTA updates to the full. It also enables real-time 3D animations and even HD video games: we expect this functionality to become more and more commonplace as the vehicle continues to evolve into an entertainment machine.
Then looking at ADAS, which also relies on a real-time 3D simulation of the real world, this will further change expectations of the digital cockpit as we move from today’s L2+ and L3 systems to the L4 and even L5 autonomy that will come in the future. As drivers transition into passengers, they’ll want to do other things, whether that’s work maybe, or perhaps truly immersive, fully customised experiences.
Today, a typical premium vehicle may contain more than 150 electronic control units (ECU) and more than 150 million lines of code – a figure expected to rise to 300 million by 2030. While the move towards central computing will help to drive down the number of ECUs and simplify E/E architectures themselves, demand for ever-more sophisticated digital cockpit technologies can only increase the pressure on hardware and software engineers. To manage all of that, the relationship between OEM and supplier must continue to evolve.
about 100 million lines of software code – four times more than a fighter jet –, projected to rise to 300 million lines of code by 2030.
The pool of pink-slipped technical workers is proving to be attractive chum for the auto industry looking to gobble up tens of thousands of software engineers and programmers as vehicles have morphed into rolling computers.That's great news for all those software engineers and other technical experts cut loose by the likes of Amazon, Google and Facebook parent Meta who might see their new futures at automakers or suppliers.
Current estimates put the number of microchips in the average car at 1,000 to 3,000. Today’s top-of-the-line vehicles have up to 150 million lines of software code
(Apple sold more than 220m iPhones last year)
Toyota sold 10.5M units in 2022, making it world’s largest carmaker
The pool of pink-slipped technical workers is proving to be attractive chum for the auto industry looking to gobble up tens of thousands of software engineers and programmers as vehicles have morphed into rolling computers.
That's great news for all those software engineers and other technical experts cut loose by the likes of Amazon, Google and Facebook parent Meta who might see their new futures at automakers or suppliers.
https://www.britannica.com/technology/iPhone Launched Jan 2007