The auto industry has seen a tectonic shift over the last decade, with software-defined transportation platforms driving connected/autonomous vehicle development emerging as the industry’s future.
According to McKinsey & Company’s Automotive & Assembly Practice, autonomous vehicles, connectivity, electrification, and shared mobility (ACES), which all rely heavily on software, were the four biggest disruptions in the automotive sector in recent years.
Software driving automotive development
The ‘2021 State of Automotive Software Development’ report released by Automotive IQ and Perforce Software in May provided vital insights into the challenges and emerging trends in this automotive industry sector after surveying over 600 automotive industry professionals around the globe.
It featured participants involved in producing a range of automotive products, including driver assistance systems, ECU/ECM, and chassis/safety systems and who worked for Tier-1/2 suppliers and OEMs. The chief concern amongst 44% of the respondents was related to the safety of automotive software, while 22% cited security as their top concern and 21% mentioning quality-related aspects of software development.
For automobile manufacturers ensuring their vehicles glitch-free during the launch is vital; however, this continues to be a challenge, with automotive software recalls rising over the past several years. Complying with the Motor Industry Software Reliability Association (MISRA) coding standards used in developing safety-critical systems has become a cornerstone of automotive development, the report states.
In standards, we trust
ISO 26262 has remained the automotive industry’s critical functional safety standard for around a decade, with 84% of survey respondents stating that they were required to comply with it. ISO 26262 covers functional safety in the event of system failures, covering Dynamic Stability Control (DSC) systems or Airbags, for example; however, it does not cover safety hazards that result without system failure.
The additional safety challenges that software developers need to manage for autonomous and semi-autonomous vehicles are addressed by the SOTIF (ISO 21448) standard, which 66% of survey participants said is a part of their software development process. A new automotive standard that has not yet come into force is ISO 21434, which focuses on the cybersecurity risk in road vehicle electronic systems. However, 71% of survey respondents stated that they would be required to comply with ISO 21434 when it comes into force; its importance can be gauged from the statistics in the report, which says, between 2019 to 2020, there was a 99% increase in cyberattacks on connected automobiles.
Quality and security are worrisome
Quality and safety concerns related to software development remain a concern for the industry, with 2019 setting a record for software-based electronic defects as per the report. “With the growing demand for electric and autonomous vehicles, this puts development teams under pressure to deliver innovative technology in shortened development cycles, the report states. With the number of software based electronic defects steadily rising, it appears however that speed in development of automotive
software is coming at the expense of quality. This will remain a highly challenging arena for automakers racing to offer their customers the maximum amount of functionality and connectivity features.
Out of the survey respondents, who cited security as their key concern in automotive software development, 46% listed unauthorised access to onboard/offboard systems as their primary concern. About 28% and 20% of survey respondents cited enforcing secure coding practices and lacking skills needed to combat security threats as other critical areas. The rapid pace of automotive software development is also putting pressure on the testing of code. The primary concern listed by survey respondents was the difficulty in enforcing coding best practices, followed by the time-consuming nature of testing and software validation, which often leaves little time for exhaustive testing when coupled with tight timelines.
An essential step to improving software quality is to measure it accurately. While there’s no one way to measure code quality, the report states that using a static code analysis tool is one of the most effective practices. With the testing process being time-consuming, the report finds that it often does not happen early enough in development, with bugs being discovered too late.
Crossing the chasm
According to McKinsey, the difference in productivity between software developers is much bigger than the one between hardware developers. Apart from the early adopters, late entrants in the automotive sector planning to grow their software capability will need to swiftly hire many engineers and create a conducive ecosystem for innovation and development. “Real change will only come when automakers update their underlying operating models for software development. Based on our research, only 40% of the R&D leaders who view software as a major disruptor feel prepared to make the necessary operational shifts,” McKinsey says.
With software complexity to nearly triple over the next ten years, McKinsey states that OEMs are awakening to the growing productivity problem and do not anticipate a quick fix. With talent in short supply, they cannot merely toss more resources at the problem. Instead, they will have to streamline software development by reducing complexity.
The automobile lifecycle further compounds the challenges for the industry, wherein a model goes between 5-8 years before a full model change and as a result will require a series of software upgrades for in-car systems, despite which it will be obsolete by the time a model change is needed, calling for sustained investments.
Image courtesy: Siemens, ARM.