Automotive prepares for the rise of cloud-first SDVs 

The emergence of software-defined vehicles (SDVs) will lead to a complete reimagining of cars and what we expect from them. Sophisticated SDVs will provide drivers with vehicles that can be updated on the go and deliver digital experiences comparable to what one would expect from any other modern consumer devices. New entrants are already making significant progress into this field, with Telsa, Rivian, BYD Lucid, and Nio stealing a march on longer-established automotive manufacturers. To keep up, traditional OEMs will need to accelerate SDV development processes.

This transition to SDV production will not be easy, but it will need to happen as the incentives are just too big to ignore.

The opportunity of the cloud-first SDV

SDVs will provide OEMs with an opportunity to retain drivers as customers after the initial sale by selling after-sales enhancements. These will be delivered not by a mechanic in a workshop but via over-the-air (OTA) updates. Rather like an app store environment, a driver will be able to buy, download and install new features. This could be a new infotainment package or different handling specifications.

It’s entirely conceivable that, instead of becoming outdated soon after leaving the showroom forecourt, cars can be improved over time, thanks to their software-centric features. In effect, they will be future-proofed. Drivers will be able to customise their vehicles and SDVs will become an extension of their broader connected ecosystem. They will be capable of syncing with other internet-of-things (IOT) products, such as a digital voice assistant.

From an OEM’s perspective, there is also the potential to design and develop automotive systems in the Cloud —and deliver software updates and fixes over the air, without needing to recall vehicles. Embracing this cloud-first approach will enable faster, more scalable, and cost-effective development and maintenance. OEMs will be able to automate more tasks and run simulations in the Cloud, independent of hardware, helping them to test concepts, hone products and bring new, improved features to market at a far greater pace.

The business case for cloud-first SDVs is clear, but the path to get there will not be easy.

The route to cloud-first SDV development

Migrating to a cloud-first SDV approach will mean moving from a long-established closed development model—based on providing a fixed set of features within a vehicle—to one that enables continuous change over the vehicle’s lifecycle. It will also mean a completely new way of developing automotive software, with much more agility and shorter development cycles.

This transition will involve significant upheaval for the automotive sector. And there will be three key steps that will need to happen before OEMs can successfully embrace cloud-first development.

To start, industry needs to commit to a centralised architecture. Automotive computing is fragmented at present. Modern vehicles often feature more than 100 electronic control units (ECUs), which manage functions such as heads-up displays and infotainment systems. This is a wasteful, prohibitively complex and costly approach. Agreeing consensus over centralised architecture will be the first key step to realising cloud-first SDV development. To achieve this, there needs to be collaboration between OEMs, software vendors, semiconductor designers and cloud innovators. Initiatives such as the SDV Alliance are working on this right now.

There also needs to be investment committed to shortening the architectural transition. OEMs will need to develop cloud-first SDVs, while still producing and maintaining legacy products. This will likely lead to them developing hybrid architecture systems as OEMs may bridge the gap to the full cloud-first SDV vision by first investing in zonal architectures and gradually migrating functions to centralised compute ECU. This will be more costly in the short term due to oversized hardware covering both the legacy and new architectures but will deliver a safer trajectory into SDVs and long-term reductions in cost and disruption.

At the same time, new automotive hardware will require built-in redundancy and regulatory collaboration. Safety is at the core of vehicle design, so any centralised architecture must ensure that the failure of any single component will not impact another, especially when crucial safety features, such as advanced driver assistance system (ADAS), are involved.

In order to avoid a single point of failure problem, this will likely result in the deployment of hardware redundancy in the centralised compute ECU, as the hypervisors can isolate critical and non-critical systems that are still susceptible to catastrophic hardware failures. In addition to this necessary redundancy, the industry must work with automotive regulators to develop new standards embracing the SDV concept and gain approval for more complex safety systems, just as we have seen in the aerospace and nuclear sectors.

Getting ready for cloud-first SDVs

While the move to a fully cloud-first environment will take time, the automotive sector cannot spend too long deliberating over the best way forward. Competition is already hotting up, with new entrants entering the race to develop cloud-first SDVs.

While the industry works towards a consensus on the standards for centralised architecture—which will allow all greater collaboration across the industry ecosystem—businesses will still need to invest. Developing interim hybrid architecture systems will come at some cost, but it will enable a more seamless and, importantly, safe transition to a cloud-first SDV development environment.

The opinions expressed here are those of the author and do not necessarily reflect the positions of Automotive World Ltd.

Przemysław Krokosz is Edge and Embedded Solutions Expert at Mobica, a Cognizant company

The AutomotiveWorld.com Comment column is open to automotive industry decision makers and influencers. If you would like to contribute a Comment article, please contact editorial@automotiveworld.com