Sjoerd van der Zwaan — Solid Sands B.V.
From Code to Confidence: Toolchain Qualification in Intelligent Transportation Systems
Intelligent Transportation Systems (ITS) increasingly rely on sophisticated C and C++ software to manage complex, real-time safety functions. While safety standards like ISO 26262 require rigorous verification of application code, the underlying toolchain — specifically compilers and runtime libraries — often receives less scrutiny. Because these tools translate high-level source code into executable behaviour, defects within them can introduce critical faults even when the source code is perfectly correct. Furthermore, the traditional approach of one-time qualification is no longer sufficient in modern environments where toolchains evolve through frequent updates, security patches, and optimizations.
To address this, we introduce continuous toolchain qualification as a systematic method for maintaining trust throughout the ITS lifecycle. This approach integrates automated verification directly into CI/CD pipelines, enabling revalidation whenever toolchains or build configurations change. This presentation details the practical implementation of such a framework, including the structure of compiler test suites, strategies for validating complex optimization passes, and techniques for scaling execution across heterogeneous hardware targets. We also discuss the role of change impact analysis and regression detection in maintaining qualification status. By establishing a reusable foundation for continuous verification, organizations can generate auditable evidence, reduce late-stage risks, and ensure sustained compliance with evolving regulatory expectations for safety-critical systems.
Sjoerd van der Zwaan is a leading expert in autonomous systems with over a decade of experience in Intelligent Transport Systems (ITS). As Chief Product Officer at Solid Sands, he directs product strategy for safety-critical components, specializing in compiler validation and toolchain qualification for embedded systems. His expertise in robotics, AI, and computer vision ensures that development processes meet rigorous safety standards. Previously, as CTO at 2Getthere, Sjoerd led the innovation of autonomous transit solutions. With a background in Electrical and Computer Engineering, he is dedicated to shaping the future of mobility. By bridging cutting-edge technology with reliable verification, Sjoerd advances smart, safe transportation solutions globally.
Yang Xiao — 2getthere
Bringing Standards into Product: ISO 26262 + ASPICE + SOTIF in an L4 Autonomous Transportation System (ATS)
Deploying safety-compliant autonomous vehicles in the real world remains a major challenge for intelligent transportation systems. 2getthere's Group Rapid Transit (GRT) system is a fully autonomous, high-capacity transport platform. In developing its collision-avoidance software, three complementary frameworks were applied in combination.
ISO 26262 provided the product-safety backbone: identifying hazardous scenarios, determining the required ASIL allocation and the corresponding development rigour per safety goal, translating abstract safety goals into concrete functional and technical safety requirements together with their architectural realisation, and consolidating the resulting evidence into a structured safety case that supports release of the vehicle.
ASPICE supplied the process-capability skeleton for software development: a lifecycle structure with defined work products, bidirectional traceability, and supporting processes. It was applied in particular to the perception software stack, so that the technical quality demands of ISO 26262-6 are met without re-inventing the development framework.
ISO 21448 (SOTIF) addressed the residual risk that ISO 26262 cannot reach: the absence of unreasonable risk arising from the intended functionality of the system or from its reasonably foreseeable misuse had to be demonstrated throughout development, and unknown hazardous scenarios encountered in the field had to be detected and mitigated rapidly to keep the function free from unreasonable risk in operation.
This presentation shares practical lessons learned from qualifying a transit system under the combined ISO 26262, ASPICE, and SOTIF framework, and offers an end-user perspective on translating safety standards and structured development processes into an operational, safety-qualified product.
Yang Xiao is the product owner for autonomous driving (AD) perception software at 2getthere, where she focuses on the multimodal sensor calibration solution and the collision avoidance solution for Level 4 autonomous Group and Personal Rapid Transit systems. Previously, as an R&D software developer, she delivered ASPICE-compliant, ISO 26262-aligned software for L4 AD projects. She holds a Ph.D. from the University of Birmingham, specialising in high-resolution automotive imaging radar for AD applications, with publications in international journals and conference proceedings. Her work bridges automotive perception research and production-grade safety engineering for autonomous mobility.
