Industry experts weigh in on SDV development challenges and how to address them. By Stewart Burnett
Software-defined vehicles (SDVs) have captured the imaginations of automakers, stakeholders, and drivers worldwide. Over-the-air (OTA) updates are at the core of the concept’s appeal, allowing features to be added or enhanced continuously without the need to upgrade hardware or visit the dealership. This, in turn, allows for the enhancement of various software functions—from advanced driver-assist systems (ADAS) to infotainment offerings—and the improvement of vehicle performance through continuous monitoring and tweaks.
However, the push to realise SDVs has not been smooth. Many automakers’ software development ventures have struggled with out-of-control costs and delays. Legacy automakers have experienced the brunt of these issues, while newer and more agile brands like Rivian, BYD, and Tesla have gained a notable technological advantage.
Automotive World consulted with several major players in the segment to hear their takes on the biggest challenges facing the realisation of SDVs and what the solutions could be.
The fluidity of the term ‘SDV’
“One of the biggest challenges is that everyone has a different interpretation of what an SDV is,” explains Arun Srinivasan, Head of Bosch Mobility UK. “What’s more, automakers have set out on different journeys to reach the SDV level that achieves their targets.” Some automakers may be satisfied with the simple ability to conduct regular OTA updates, whereas others may opt for a more radical approach.
This is corroborated by Mario de Felice, Head of Software Architecture at Jaguar Land Rover (JLR): “Everyone has a different vision of SDV.” JLR’s own version of the concept is a “fully-tailored experience” for customers that spans both in-vehicle services and infotainment offerings, as well as connectivity to other user devices and the wider environment, strong cloud connectivity, and deep AI integration.
Moritz Neukirchner, Senior Director of Strategic Product Management for SDV at Elektrobit, believes that greater clarity around the concept could be realised through a common terminology tiered from 0 to 5. Level 0 would entail the most basic software functions like parking assist and adaptive cruise control, while Level 1 would integrate connectivity with the user’s smart device or live traffic monitoring. At Level 2, OTA bug fixes become possible, and Level 3 OTA would allow for new features to be added.
The upper two tiers have a more profound effect on an SDV’s overall value proposition. Level 4 turns the vehicle into a software platform in which the lifecycles of hardware and software become decoupled. Finally, Level 5—which no automaker has yet achieved—would see the vehicle become an open-source innovation platform, offering the driver a wide range of digital products and services from third parties. “This system really helps parties to talk clearly about their ambitions,” explains Neukirchner. “The SDV value proposition will not come in all domains to the same degree simultaneously.”
Legacy mindsets persist
While a lack of mutual understanding about what an SDV really is can make collaboration unnecessarily cumbersome, Neukirchner believes a bigger problem lies in how software development is considered at the organisational level. “It’s the general approach—the business and purchasing processes are not conducive to lifetime support,” he explains.
In the past, all automakers had to do was produce a vehicle at a factory, and Neukirchner worries that this legacy mindset persists. “Many OEMs are still treating software sourcing as something that contributes to the cost of production and is worked on up to a certain point. The purchasing process is treated as a one-time thing, which is fundamentally mismatched with the reality of an SDV.” Treating software development as one-time expenditures, he emphasises, causes development costs to balloon.
Many automakers have turned to partnerships, something Christoph Horn, Global Lead for SDV at Accenture, deems a positive development. “OEMs have learned the hard way that to get speed at the right cost, they need to partner.” He believes the solution is to make these partnerships long-term arrangements. Keeping the same people involved improves mutual understanding of the automaker’s software platform and makes it easier for products and OTA updates to be deployed across a vehicle’s entire lifecycle.
However, this entails a massive and continuous workload. Horn suggests using AI to ease the pressure: “We expect AI will be very dominant in the next few years. It will definitely be used to speed up software development around vehicles and reduce all the complexity we currently see.”
Customer expectations are changing rapidly
For Accenture, the big challenge is simple: addressing new developments quickly and efficiently. “Customers are shifting their expectations dramatically,” explains Horn. “This means that the traditional unique selling points of brands are vanishing, and you need something new. Start-up OEMs have been very fast in observing, addressing, and even driving this shift.” De Felice and Srinivasan both state that the pace of change is aggressive. “We need to make sure we maintain speed in delivering updates and fresh experiences for our customers,” de Felice notes. “The overall process of software creation, validation, and delivery needs to accelerate so JLR can deliver a unique modern luxury experience.”
From what I’ve seen, OTA performance has been rather poor across the board
JLR intends to address changing customer expectations by implementing an efficient continuous integration/continuous delivery pipeline and increasing the amount of automation in its processes. “We’ve invested heavily in our vehicle connectivity and data platform, which manages data, runs applications and delivers services online to create a tailored and fully personalised experience.” For Srinivasan, scalable solutions are the key: “These are best co-created between OEMs and their partners. It’s about having real fluidity to be adaptable and create assets together that can be scaled, re-used, and built upon.”
At the same time, Neukirchner emphasises that the rush to deploy new features cannot compromise their quality. “From what I’ve seen, OTA performance has been rather poor across the board”. He identifies two flaws: first, customers may not even perceive that a new feature has been implemented; and second, these updates can cause unforeseen glitches and even break the vehicle in some cases.
The consensus is that agility must not come at the expense of quality, and striking this balance means having the right processes in place from the outset. Indeed, this approach might be able to address all aforementioned challenges. “The vision must start with software itself, not with putting software in a car,” Horn concludes. “Naturally, people always feel inclined to do what they’ve done before, but this is a wholly new field of expertise. How to define, develop, and scale software is determined massively by the processes you establish on day one.”