How Schaeffler AG accelerates automotive software development by leveraging the AWS Virtual Engineering Workbench Framework

Introduction

This blog post will detail how adopting Virtual Engineering Workbench (VEW) has supported Schaeffler AG, an automotive supplier, within their software development for pre-series development projects. It will also explain the use cases implemented by AWS and Schaeffler, and present one of the many technical examples. This includes how Schaeffler was able to adopt VEW in order to help them:

• Reduce time needed to set up new virtualized development environments from 30 mins to 6 mins.
• Reduce time needed to set up new tools for pre-development environments from 2 days to 45 mins.
• Enable automated validation by integrating virtual targets (virtual versions of electronic control units) into CI/CD pipeline.
• Optimize the developer connections to virtual systems to 15ms on average.

First some context: In today’s fast-paced automotive industry, companies developing software for Software-Defined Vehicles (SDVs) face significant challenges. Electronic Control Unit (ECU) development currently depends on afterwards testing software on the automotive grade hardware, which results in extended time-to-market, inefficiencies in the development process, and difficulties in maintaining quality and reliability across complex systems. For Automotive suppliers, the typical product development lifecycle for hardware and software systems spans multiple years, encompassing stages from initial requirements gathering to pre-series production.

The Schaeffler Group has been driving groundbreaking inventions and developments in the field of motion technology for over 75 years. Through the merger of Schaeffler and Vitesco Technologies emerges a leading motion technology company, that makes motion more efficient, intelligent, and sustainable. Schaeffler’s claim sums it up very well: “We pioneer motion”.

Virtual Engineering Workbench is a framework, guidance & AWS implemented accelerator that brings developer tools & various versions of virtual / physical targets in one self-service portal for developers. It allows users to spin up hundreds of test environments in parallel but also enables developers to rapidly test their own changes.

Using VEW to address software development challenges

VEW is a cloud-based development environment for embedded software development. It is built on four core principles: self-service, global scalability, reduced hardware dependency through virtualization, and the creation of consistent and reproducible software artifacts. VEW aims to enhance software developer efficiency, business agility, and developer satisfaction by providing pre-configured yet customizable environments, utilizing virtualization technologies, and implementing a high degree of automation.

By supporting the product design and development stages, from software architecture to system testing, VEW offers Schaeffler users the “right environment for the right job”. VEW provides pre-configured developer environments for a specific project task which makes tool management & version management simpler. Additionally, VEW helps streamline collaboration with other suppliers and enables efficient sharing of environments & targets.

By leveraging “virtual control units”, which are virtualized environments of future car components, developers can validate changes quicker in a self-service fashion which helps avoid time delays by reducing feedback loop timing. These same virtual control units can be used for automated testing in order to run tests more often and find potential errors earlier.

How Schaeffler was able to leverage VEW to accelerate their projects

The project team took an architecture, of an Innovation project, that leverages the AUTOSAR Adaptive standard. AUTOSAR Adaptive, is a software standard designed for High Performance Computing automotive applications that offers standardized interfaces and services for applications. AUTOSAR Adaptive supports complex functionalities such as real-time data processing and safety relevant functions. This is essential for the powertrain domain (everything inside the car that transmits drive from engine to wheels) that Schaeffler is focused on.

For this use case, we identified all the tools required to develop an application. Next, for Schaeffler developers VEW is a pre-configured versioned cloud-based integrated development environment. The developers could then provision AUTOSAR Adaptive workbenches on-demand using VEW which enabled them to develop and test software within as little as minutes. The VEW manages the lifecycle of the workbenches completely. The VEW brings together software developers and workflows for managing their environments in one solution. This runtime environment then is integrated into Schaeffler’s CI/CD (Continuous Integration/Continuous Delivery) pipeline. SDV development in the automotive industry requires the use of complex tool chains involving tens of software tools. To help ensure safety of the car passengers and pedestrians, Original Equipment Manufacturers (OEM) must implement high safety and quality standards of the produced software components. A software development tooling solution allows for reproducible development environments for years into the future. Another important concept is tooling parity. This means that the development environments must match exactly with the build environments used in the CI/CD pipelines. This includes having identical versions of all tools and software. It’s also crucial to have a process for versioning these tooling bundles. In automotive software development, this approach is vital for several reasons: 1/ consistency across all stages of development, 2/ transparency in the development process, 3/ traceability between development, build, and test environments 4/ faster feedback on changes made to the software. By maintaining this tooling parity, automotive software developers can more easily meet industry standards for quality and reliability, while also improving their development efficiency.

Architectural Deep Dive: VEW Integration to Schaeffler CI/CD pipelines

Continuous Integration (CI) and Continuous Delivery (CD) pipelines are essential to modern software development lifecycle (SDLC) process, and automotive SDV development is no exception. SDVs have multiple Electronic Control Units (ECUs), each of them having a dedicated repository to version the source code. Each of these repositories can have a different CI/CD pipeline setup, where the build image could be created with different tooling versions or toolchains. Developers may not even have control of which versions are used for a particular build and obtaining the required version of the tools needed to build the software can be a lengthy process. For example, at Schaeffler where for innovation projects the CI/CD flows are detached of developers’ individual environments and this can lead to gaps & inefficiencies. Lastly, these tools can have dependencies to specific versions of other required tools causing more management struggles. VEW helps customers solve these problems by allowing developers to launch a cloud-based workbench from anywhere with the required and tested tooling version, which is identical to the one used in the automated build environment, in as little as 5 minutes.

AWS and Schaeffler integrated VEW to GitHub Actions as self-hosted runners. This setup enables Schaeffler developers to use modern CI/CD tools for automotive development. Workbench versions that VEW contributors’ package is also available in a GitHub Actions workflow. Developers can reference the workbench version as a custom runner in the workflow definition.

Figure 1 illustrates how AWS and Schaeffler integrated VEW with GitHub using VEW-packaged Amazon Machine Images (AMIs) as base for the runners.

Figure 1: GitHub Runner workflow

The VEW runs the workbench runner workflow as following:

1. Product contributors package an AMI using the VEW packaging feature
2. Product contributors publish a new product version in the VEW product catalog, selecting the AMI from step 1
3. The VEW publishing engine shares the AMI with the AWS accounts that will host the workbenches and the AWS accounts that will host the GitHub runner machines.
4. After sharing the AMIs, the VEW updates the configuration with new labels of the GitHub Actions runners with the newly published product version
5. GitHub provides the GitHub Actions runners configuration to the pipeline accounts
6. Schaeffler developers configure a GitHub Actions workflow in their application repository with a pinned product version label, for example AUTOSAR-adaptive-1.0.0
7. Schaeffler developers provision the corresponding product version in VEW. A developer can now clone project’s code repository and starts contributing. The combination of the shared product AMI and the GitHub runner configuration, ensures the CI/CD pipeline build the artifacts with the same AMI that the developers used to code new features or big fixes.

Conclusion

The project took AWS and Schaeffler 12 weeks to implement, during which the combined team successfully collaborated and enabled Schaeffler engineers to develop, build and test software for AUTOSAR Adaptive in Innovation projects on demand using AWS. The AUTOSAR Adaptive workbenches are pre-configured, licensed, and versioned which helped ensure ease of use and consistency. With the end-to-end integration completed, the Schaeffler & AWS team created consistent, reliable and high-performing development, build, and test environments which enabled faster feedback cycles and improved developer efficiency.

We measured the success of the project with three metrics:

1. Improved developer efficiency: Schaeffler reported a reduction in time to setup a virtual development environment for their innovation project from 30 minutes to as little as 6 minutes with VEW. In addition, Schaeffler reported that VEW improved the time needed to install all required tools/licenses from 2 days to 45 minutes.
2. Improved business agility: we enabled the use of virtual targets with required tooling installed in CI/CD environments which helped shorten feedback cycles.
3. Improved developer satisfaction: The project team held interviews and optimized the connection so that we have best user experience and provided them a virtual environment with as little as 15ms latency.

Building on their current successes, AWS and Schaeffler plan to further improve VEW in two main ways:

1. Expanding VEW’s capabilities: They aim to increase the range of use cases that VEW can handle. This will make the tool more versatile and useful for a wider variety of development tasks.
2. Integrating advanced technologies: They plan to incorporate Generative AI into VEW. This addition will enhance VEW’s ability to assist with requirement analysis and testing.

Additionally, AWS and Schaeffler will focus on getting more developers to adopt and use VEW in their work. This increased adoption will help maximize the benefits of the improved system across the organization.

Your needs, objectives and circumstances may be different and may need to be addressed differently. Please reach out to AWS Sales, AWS Professional Services, or to the AWS Partner Network to discuss your optimal solution. If you want to learn more about how Virtual Engineering Workbench helped AWS Customers please also check this Blog post and this Video from Stellantis.