Developing ISO 26262 and SOTIF–Compliant Structured Text for Low-Speed Autonomous Vehicles

Supported by IFM Electronic, Project Helps Develop Industry-Relevant Skills for Students

“We chose MATLAB and Simulink to implement our safety layer algorithms [because] they come with an ISO 26262 certification toolkit and allow for static [checks] on both the model and generated code.”

Key Outcomes

Rapidly prototyped control algorithms for autonomous vehicles, enabling testing in scenario-driven 3D environments
Generated certifiable ISO 26262 and SOTIF–compliant structured text using MATLAB and Simulink
Developed teaching materials for embedded control system design from project results, reflecting the latest industrial topics of interest
MATLAB and Simulink tools enabled the creation of an open-source, reusable safety concept backed by industry standard–compliant processes and documentation

A diagram of HAN’s workflow for implementing the safety layer, comprising IEC Certification Kit for ISO 26262 and static code and model analysis. — Workflow for the AV safety layer using static model/code analysis and formal verification.

Hogeschool van Arnhem en Nijmegen (HAN) is a Dutch university of applied sciences that bases its curricula on research conducted with industry partners. As a lead partner of the Safety Concept for Low-speed AV Implementations (SafeCLAI) project in the Netherlands, HAN, with support from IFM Electronic, designed a reusable ISO^® 26262 and SOTIF–compliant safety layer for ROS-based autonomous vehicles.

Choosing MATLAB^® and Simulink^® to implement the safety layer gave HAN access to IEC Certification Kit, enabling standard-compliant verification and validation workflows for generating certifiable and integrable code. ROS Toolbox helped the HAN team overcome the challenge of transferring data from simulated sensors in the 3D environment through the use of specialized ROS messages. IFM Electronic supported the project by providing expertise in functional safety and supplying a safety controller executing the generated structured text in a safe controller architecture.

As a result, HAN now has a reusable, scenario-driven verification and validation environment in which to safely test autonomous vehicle systems. This workflow is expected to continue growing and become better integrated in the future. For instance, because ROS lacks workflows that can be certified, HAN is considering using RoadRunner for simulated driving scenarios and moving the workflows for image and signal processing to MATLAB and Simulink. For future projects, moving to generated C code and adopting Polyspace^® is also being considered to check and prove MISRA™ compliance of the integrated code and to enhance toolchain consistency.

The project’s source code and related documents have been published on GitLab^® for industry use and educational purposes. Additionally, HAN’s curriculum will be enriched with course material on functional safety from the SafeCLAI project, helping students develop valuable employability skills by exposing them to how code generation tools, along with MATLAB and Simulink, are being used in the industry.