Stage

Master 2 Internship - Combinatorial testing for autonomous vehicles

Équipes / Services concernés

Responsables

Hélène Waeselynck / Jérémie Guiochet

Date de publication

03.11.26

Prise de poste souhaitée

01.02.27

Master 2 Internship Subject - Combinatorial Testing

Title : Combinatorial testing for autonomous vehicles
Supervisors : Jérémie Guiochet (homepages.laas.fr/guiochet), Hélène Waeselynck (homepages.laas.fr/waeselyn/)
Profile : Computer science
Level : Master 2 (Msc)
Duration : 5-6 months
Location : LAAS-CNRS, Toulouse, France (www.laas.fr)
Salary : ~600e/month

Producing test cases (or test scenarios) to validate a system under test has been a major challenge in the testing community for decades. This challenge has become even more important with the emergence of autonomous vehicules, which can be evaluated using high-fidelity 3D simulators. Test case generation and selection pipelines have been extensively studied and developed in many domains, including automotive systems and robotics. They generally follow a common structure, starting from a high-level test specification and producing concrete test cases, i.e., the data injected into the system under test (e.g., the software of the autonomous vehicle). However, the choice of which scenarios should be selected for performing the test is still an open challenge, and different strategies are studied. Among them, the t-way selection, based on a combinatorial exploration of the parameters values (when t=2, it is called the pair-wise testing), is of high exploration but still limited to simple test case structures.

The objective of this internship is to investigate how t-way combinatorial testing can be integrated into TAF (https://wp.laas.fr/taf), a framework developed at LAAS-CNRS for generating structured test cases under constraints. Although combinatorial testing has already been applied to autonomous-vehicle testing, its integration into a TAF-based workflow raises specific questions about parameter selection, constraints, and coverage of complex test specifications. These questions will be investigated through a case study involving the generation and selection of test scenarios for autonomous-driving software evaluated in CARLA (https://carla.org/).

Application : CV + cover letter + transcripts of records for the last two academic years