If software is eating the world, it is devouring the automotive industry. But many long-standing vehicle makers are not organized to develop and test software quickly and efficiently. That needs to change—and fast. Otherwise, new entrants—such as Tesla and BYD—will take even bigger bites out of incumbents’ market share.

Automakers have to think and act like software companies. The development of software-defined vehicles is driven by the need for faster product releases, advanced features such as ADAS, and enhanced connectivity, as well as pressure to manage costs and be competitive with new entrants. Yet traditional roles, processes, and organizational structures struggle to keep pace with the need for rapid innovation and integration, leading to inefficiencies and missed market opportunities.

But if they apply the right organizational principles and approach, such as hiring and empowering skilled integration engineers, traditional OEMs could reap major benefits. They could, for example, achieve a fivefold increase in the speed of software verification and validation (V&V), through a dramatic reduction in process complexity involving a fourfold reduction in the number of different roles. Speed isn’t the only benefit: software quality will rise as valuable testing outcomes (in terms of findings and coverage) increase by between 10 and 20 times.

How can an automaker achieve these gains? Adopting the agile methodologies and CI/CD (continuous integration and continuous delivery/deployment) mechanisms that are commonplace in the software industry will accelerate development cycles and improve market responsiveness. To maximize product reliability and functionality, it is vital to seamlessly integrate software and hardware. That can be achieved by building a collaborative culture in which cross-functional teams and open communication drive innovation and faster problem-solving. At the same time, automakers need to utilize state-of-the-art tools to streamline workflows, improve quality, and reduce development spending. In this way, they should achieve a major reduction in costly recalls, which are a serious drag on profitability in the automotive industry.

By applying these principles, Tesla is outpacing most of its competition. Tesla takes just 12 hours to get results from the first stage to full-vehicle testing, compared with between four and six weeks for a traditional OEM. That’s because it performs V&V integrations fully automated and in parallel, rather than taking a sequential approach. At Tesla, just four people are required to test a feature, while three or less roles are involved in validation. The equivalent figures for a traditional vehicle maker are 42 and eight. As a result, Tesla has just 90 people working in its software quality department, whereas an incumbent automaker might have more than 300 (see figure). In addition to being more cost-effective, a smaller software team is more agile.

By analyzing the approach taken by Tesla and other truly software-driven automakers, BinaryCore and tracetronic have identified five key lessons for traditional players.

1. The pivotal role of the integration engineer
2. Responsibility for V&V spread across the different stages of the process
3. Parallel processing of tests performed
4. The importance of automating regression testing
5. A central and transparent results database created

The pivotal role of the integration engineer

Firstly, let’s consider the pivotal role of the integration engineer in driving a feature from development to release. This dedicated role is designed to transcend traditional compartmentalization and help streamline the organizational structure, increasing efficiency and cutting down process time and the length of a V&V round trip.

The most important person in software V&V, an integration engineer steers software feature development and validation. As they are responsible for taking the software through all the validation stages, they need expertise in integration, mechatronics, project management, and communication, as well as leadership skills. This combination of skills and talents can be hard to find, requiring a sophisticated approach to recruitment.

Responsibility for V&V spread across the different stages of the process

The second of the five lessons is the importance of streamlining the interfaces and reducing the handoffs between different stages of the V&V process. By putting a focus on integration, rather than function, the goal is to ensure that work on each single phase is perfectly aligned to the requirements of subsequent stages. The result is less redundant testing and a more effective workflow. To that end, automakers should look to expand and empower the role of the validation engineer to cover more areas and to take on more responsibilities. They need to be able to take the initiative, make decisions, and drive improvements across the different phases of the V&V process.

Parallel processing of tests performed

Another key takeaway is the value of parallel processing. By leveraging smart methodologies to execute tests in parallel, automakers can accelerate the production of high-quality software fivefold. To realize that acceleration, there is a need to remove the barriers to testing, automate each process as much as possible, and ensure that developers receive feedback quickly from the validation engineer. Rather than making developers jump through multiple hoops before testing software, they should be encouraged to “dare to fail.” Transparency is key to moving quickly, so errors should not be punished.

The importance of automating regression testing

The next critical step is the automation of regression testing, overseen by an agile and highly skilled validation team. Regression testing ensures that software, which has been modified or updated, still works as intended. This approach requires extensive usage of APIs, machine-processable test management, and even the development of proprietary tools to achieve the required levels of automation, as well as early engagement of all validation stages in feature development. Indeed, configuration management is key to ensure software and hardware compatibility.

A central and transparent results database created

The final lesson is to establish a central and transparent results database. In addition to increasing the speed of software development, this database provides flexibility, enabling efficient portfolio management. It also allows for real-time monitoring, which enables a quick response to emerging bugs. As a centralized solution with no external bias, the results database provides both data contributors and consumers with full visibility of the current status of all software under development.

In summary, long-standing automakers still have a great deal to learn about software development from the approach taken by Tesla and other newcomers. It is possible to increase both speed and quality by streamlining processes, reducing the number of interfaces, and empowering key employees. In this way, high-tech newcomers verify and validate software far faster than traditional OEMs. As a result, they can bring new vehicles and features to market faster and more efficiently. They have both a cost advantage and a quality advantage. All automakers now need to master software V&V.

tracetronic

tracetronic is a leading software product provider for consistent sw testing with leading-edge agile product development processes. With a focus on the automotive sector, the product portfolio combines both intuitive usability and technological depth. It flexibly adapts to agile development workflows, handles dynamic changes with ease, and delivers experienced users the tools they need.

For more info visit tracetronic GmbH.