Applying Proven Robotics Techniques to Automobile Development Processes

SOA allows for the decomposition of automotive systems into individual services that communicate over a network, enabling more scalable and reusable components. This is particularly crucial as vehicles become more like “computers on wheels,” requiring rapid development cycles and greater flexibility to incorporate new technologies or updates without overhauling the entire system.

The increasing complexity of modern vehicle systems, with numerous sensors and actuators that must operate cohesively, is driving Renault’s transition to an SOA. Adopting an SOA allows Renault to develop and test different vehicle components independently while still ensuring they function seamlessly together.

To facilitate this transition, Renault developed a system that allowed for rapid and independent testing and prototyping well before implementing anything on hardware. Simulink^® products helped Renault streamline its work with software units and its composition into full applications while considering the shift toward an SOA-based future.

The integration team, led by Pierre Muratory, focuses on deploying Simulink models onto various targets, including ROS-based prototyping platforms and the production AUTOSAR platform, which is the global standard for software architecture in vehicles. The team conducts continuous integration and continuous development (CI/CD), automatically updating the pipeline for each target whenever they make a change.

“The advantage for developers using Simulink is that it is totally agnostic to the platform and the target,” Muratory says. “We use Simulink as a cross-platform tool to deploy the same code source from the Simulink model onto different platforms.”

This integration flow powered by Simulink, a software factory, CI/CD pipelines, and automated cross-platform ROS code generation allows Renault to quickly transition from a developer’s model release to running the integrated code for vehicle prototyping and testing.

“Our continuous integration flow using Simulink and ROS Toolbox allows us to go from a developer’s release to prototyping and deploying on real targets in under an hour,” says Muratory.

Significantly reducing cycle times from days or weeks to just an hour is a game changer for Renault’s ADAS prototyping and early feature validation efforts. This accelerated process, enabled by automatically generated code from Muratory’s integration team, allows engineers to quickly spin up ROS nodes to test cutting-edge capabilities well before production rollout. Pascal Bedouet leads a team that takes full advantage of this agility, focusing on leveraging the ROS infrastructure for advanced communication and ADAS prototyping.

Renault also adopted ROS 2. It’s an upgraded version of ROS, which Open Robotics, the nonprofit that supports ROS, plans to stop supporting in the near future. ROS 2 provides significant advantages for automotive design. It supports real-time operations crucial for autonomous driving functionalities and includes advanced middleware capabilities that facilitate more reliable data distribution between the vehicle’s interconnected systems, something not possible with the earlier version of ROS.

“Now with ROS 2 and new middleware like DDS (Data Distribution Service), we are able to reach and meet real-time performance today,” Bedouet states.

MathWorks supported Renault in the transition, including working closely with the engineering team to incorporate functionality in the toolbox to work better with SOA. That included creating more custom messages for the different components of a system, designing a Simulink ROS 2 server block, and meeting Renault’s specific real-time requirements. Through close collaboration with MathWorks, Renault overcame these challenges without adding significant delays.

“Embracing SOA is a strategic move for Renault, and the critical support for ROS 2 service servers in Simulink has been instrumental in advancing our SOA application,” Muratory says.

Renault’s feedback helped guide MathWorks updates on ROS Toolbox, which will benefit other customers. According to YJ Lim, a robotics product manager at MathWorks, this is especially important as more automotive companies are turning to ROS 2 for their ADAS development.

A core advantage of using Simulink combined with ROS Toolbox is that developers can focus purely on their algorithms rather than dealing with integration details or specifics of the ROS framework. The Simulink models containing their design are automatically transformed into deployable code for the different targets through the continuous integration workflow setup by Renault’s integration team using MATLAB^® and Simulink.

“It will be great to achieve certifications such as ASIL for safety, which we cannot do today without a certified middleware,” Bedouet said.

By pioneering the integration of SOA principles with the ROS framework and Model-Based Design using Simulink, Renault is establishing itself as an industry leader for modular, updatable, capability-rich, and safety-certified automotive software systems for tomorrow’s intelligent vehicles.

“The support from the MathWorks team has been instrumental in enabling us to deploy and test our next-generation ADAS features ahead of schedule,” says Bedouet. “This agility has been a cornerstone in successfully introducing new capabilities in our vehicles.”