Introduction to Variant Testing and AI

[Symphonic Beat] Welcome to this edition of Test Talk, where we discuss the latest developments in science and software testing with leading practitioners. Today, we’re joined by Dr. Mehrdad Saadatmand, a leader in variant testing, to discuss the application of AI and variant testing with Bombardier Transportation, part of the Alstom group.

Background of Dr. Mehrdad Saadatmand

Dr. Saadatmand is a senior researcher at RISE Research Institutes in Sweden, specializing in model-based engineering of real-time embedded systems. He has led various international research projects, including CHESS, SAVE, and the MBAT European Project, focusing on model-based analysis and testing of embedded systems.

Introduction to the XIVT Project

Mehrdad and I met through the XIVT project (eXcellence in Variant Testing), which addresses the challenges of variant testing in complex, configurable systems. The project involves five countries: Canada, Sweden, Germany, Turkey, and Portugal, with Mehrdad’s team leading the Swedish consortium.

Mehrdad’s Involvement in Variant Testing

Mehrdad’s research at RISE centers on developing verification and validation methods for cyber-physical systems. His work is crucial for companies like ABB Industrial Automation, Bombardier Transportation, and Volvo, which rely on resource-constrained systems requiring rigorous testing.

Importance of Variant Testing for Industry

Variant testing ensures product quality across different regional standards. Companies developing products for diverse markets must ensure compliance with local standards, making variant testing essential to meet global quality requirements.

Optimizing Product Variant Testing

In variant testing, a key question is how to select which product configurations to test under tight time-to-market constraints. Optimization methods can help identify the subset of configurations needed to ensure quality while delivering products efficiently.

Role of Manual and Automated Testing

Manual testing, though valuable, faces scalability challenges with complex products. Automated testing solutions, particularly for generating test cases for new or modified features, have become necessary.

Using AI and NLP in Variant Testing

Mehrdad’s team applies machine learning and natural language processing (NLP) to analyze new product requirements and identify reusable components. Their work with Bombardier reduced analysis time from weeks to minutes, automating the reuse of components for new product versions.

VARA Tool for Variant Testing

Mehrdad’s team developed the VARA tool (Variability-Aware Reuse Analysis), which uses machine learning and NLP to recommend reusable components across projects. This tool enables companies to rapidly identify components for reuse, optimizing variant testing and product delivery.

Future of Variant Testing Research

As software-intensive systems grow, the future of variant testing lies in further automating and optimizing testing processes, leveraging AI to identify testing priorities and patterns in issues across product versions.

Conclusion and Contact Information

To learn more, visit xivt.org or contact Dr. Saadatmand directly for specifics on VARA and RISE’s involvement. Thank you, Mehrdad, for this insightful conversation and your leadership in variant testing.