Software’s Hard Challenges

Several companies at the dSPACE World Conference detailed ways that software-defined vehicles are forcing them to alter design practices.

Example of a simulation architecture for testing closed-loop trajectory planning from a vehicle’s sensor data. (dSPACE)

Automated driving, electrification, cloud computing and the push toward software-defined vehicles are forcing automotive and commercial-vehicle developers to revamp design strategies. Tools suppliers are moving to help engineers develop and verify solutions that address the complete vehicle environment, a task that requires a growing number of design tools.

dSPACE CEO Carsten Hoff detailed plans to see if some companies can work together to help users share data created using disparate design tools. (dSPACE)

During the recent dSPACE World Conference in Munich, Germany, several vehicle manufacturers described their strategies for coping with these trends. dSPACE, which supplies hardware/software-in-the-loop (HIL/SIL) tools, announced plans to see if tool makers can find a way to help developers by making it easier to integrate data created using different development software.

“Several companies are doing verification and validation tools, so it’s a challenge to integrate everything into a seamless tool chain,” said Carsten Hoff, CEO at dSPACE. “We want to invite companies to discuss creating an industry platform. We want to see if we as an industry can find the right consortium, with different companies contributing to make life easier for users.”

dSPACE’s SIL platform is open to integrate all kinds of models, virtual ECUs and third-party components. (dSPACE)

He explained that a group of 10 or more companies, possibly a mix of tool suppliers and system developers, could develop common platforms or interfaces that would help engineers and programmers integrate data created using different tools. The need for simplified data sharing was echoed by other speakers at the conference.

“The industry needs clear, unambiguous exchange of data between tools. We need standards people can trust, and standards that can evolve rapidly to stay abreast of technical advances,” said Benjamin Engel, CTO at ASAM, a standards body.

Digital continuity

“Clear, unambiguous” data sharing is a goal of ASAM’s Benjamin Engel. (SAE/Terry Costlow)

As complexity increases, there’s a strong push to ensure that data developed in one area can be made useful in others. A strategy that goes from design to manufacturing to over-the-air (OTA) software updates is becoming a necessity. That dovetails with the growth of design software that addresses many parameters, such as software, hardware, mechatronics, testing and manufacturing.

“Companies need to be able to use cloud technology to link all the players together, said Udo Lange, VP digital engineering and R&D, Capgemini Invent, a consulting company. “Digital continuity means not just data collection – companies need to integrate data for use in different areas like manufacturing. They need to integrate everything to their back-end processes, extending to distribution of OTA updates that are managed well and distributed accurately.”

BMW is making standards a centerpiece of its efforts to reduce complexity; AUTOSAR and Ethernet are standards that are seeing increased use at the automaker. Employing standards helps in validation and verification as well as in basic product design.

“The only way to reduce complexity is with harmonization and standards,” said Kirsten Matheus, senior expert at BMW Group. “Without standards, we don’t get the right tools, and complexity rises.”

Several conference speakers noted that creating and organizing virtual and physical tests often is a challenging task. As design and testing tools offer more functionality, they become increasingly complex. Users, however, are pushing for systems that are simpler to employ. Toyota Motor Corp. recently switched from PCs to dSPACE systems for its SIL testing. With that switch, a standard graphical user interface (GUI) was created.

“One of the most important requirements for our SIL platform is usability,” said Yuya Nagasawa, simulation engineer at Toyota. “Toyota created a GUI that combines all our processes into one user interface. Visual Studio, Matlab and others all use the same GUI.”

New design strategies

Designers are changing their strategies to cope with the expanding growth of software in commercial vehicles. (Hyundai)

Stefan Schmerler, head of E/E Pre-Integration Autonomous Driving and Body at Mercedes-Benz, described some of the techniques being used to simplify the task of creating reliable vehicles that increasingly rely on software for critical features and functions. Since 2018, the company has made dramatic changes to address the growth of software.

“We changed our whole research and development organization, and our process methodologies and tools have changed,” Schmerler said. “We standardize over all domains, that’s very important due to complexity. In most cases, we speak to AUTOSAR. When we don’t have it, we usually have big troubles.”

Mercedes-Benz has “changed its whole R&D organization, Stefan Shmerler told dSPACE conference attendees. (SAE/Terry Costlow)

Both real-world and virtual testing are mainstays of Mercedes’ new approach. HIL clusters increasingly are used to test concepts and find faults before they become embedded in designs and production vehicles. Testing’s importance is rising. “Design for testability is one of the biggest tasks in the future for electrical and electronic architectures,” Schmerler said.

Commercial-vehicle designers also are changing their strategies to cope with the expanding growth of software. Martin Zeilinger, head of Commercial Vehicle Development at Hyundai Motor Co., stated that more and more data is flowing through vehicles, often with time-sensitive requirements. That’s prompted Hyundai to shift to a zonal electrical platform in which groups of processors and sensors are combined to feed data to a centralized controller. That simplifies the architecture and reduces communications, but these changes require a new design strategy.

“It will be a paradigm shift for us to a software-defined vehicle strategy,” Zeilinger said. “Functionality will be driven by software. We’ll use more HIL to test and move to more standardization in our tool chain.”

Software-defined CVs

The explosion of software is not limited to passenger cars, which could see software soar from around 100 million lines of code to 300 million by 2030. Commercial vehicles also are becoming software-defined vehicles.

Strict adherence to scheduling is among Daimler Trucks’ strategies for managing this growth of software. The company currently has two major releases each year, with seven so-called “sprints” for developing new features. “It’s mandatory to use release management for all software releases. If you don’t have a mandatory release schedule, management can become a nightmare,” said Thomas Bardelang, manager of high-voltage systems at Daimler Truck.

The complexity of new vehicles extends to systems that often are overlooked in discussions of overall vehicle performance. Reduction in vehicle noise levels poses a challenge for those who design smaller systems in vehicle cockpits. Philipp Hugelmann of Robert Bosch said that using HIL made it possible to dramatically cut noise levels for equipment such as window lifts, wipers and seat drives.

“In one test, we reduced noise by 10 dB. The event went from a very annoying noise to a noise you could barely hear. After setup, we needed only three hours for this optimization,” said Hugelmann, a system-validation engineer at Bosch.

Rise of AI

Artificial intelligence is becoming more important in the development and management of vehicle software. Jürgen Bortolazzi, head of driver assistance and automated driving at Porsche, said it offers much potential for both end products and in vehicle design.

“AI has huge potential in process development and our tool environment,” he said. “We already have some AI tools. We’re using Mobileye software and AI for some ADAS systems.”

University researchers are exploring ways that AI and neural networks can help vehicles adapt to changes in their operating environment. Neural networks can allow systems to change operating parameters as needed, according to Johannes Betz, professorship for autonomous vehicle systems, Technical University Munich. His team used HIL to develop open-source software that can be used to develop powertrain software that can respond to oil spills on roadways, for example.

Charging forward

Technologists working on electric vehicles (EVs) also are dealing with rising complexity. There is a range of standards for charging, posing challenges to ensure that a base design can operate with different charging stations. Newer chargers may include bidirectional charging and technologies that let drivers automatically pay when they plug in.

Andreas Aumann, VP of strategic product management for accessories at BMW Group, noted that standards like ISO 15118 for vehicle-to-grid communication for EV charging are “the right way to go.” He added that BMW is confident that half its sales will be EVs by 2030.

As the charging infrastructure grows, much of the transportation industry’s work will be to eliminate software glitches. Most hardware issues have been resolved, but many software issues remain, according to Accenture’s Sabine Hug, manager of EV charging. She described problems such as some systems that ask for information before vehicles are plugged in, while others reverse those steps. That’s extremely confusing for users. Conformance testing by third-party organizations is needed to avoid this type of issue, but it’s only one step in the testing process.

“Conformance tests only cover one software standard, not the entire chain,” Hug said. “We need end-to-end tests. Things get very complex when standards add new sequences and features like bidirectional transfers.”