AI-Assisted Variability Analysis and Configuration of Robotics Software Systems

Robotics software systems are large, distributed, and highly configurable. A single robot platform may support multiple hardware configurations (e.g., sensors, actuators, degrees of freedom) and deployment contexts, leading to significant software variability. Today, this variability is managed manually through configuration files, conditional logic, and ad-hoc architectural decisions, making robotics software difficult to understand, configure, and evolve.

This project explores how Artificial Intelligence for Software Engineering (AI4SE)—in particular Large Language Models (LLMs)—can assist developers in analyzing, reasoning about, and managing variability in robotics software systems. Robotics software (e.g., ROS-based systems) serves as a realistic and challenging domain for studying AI-assisted software analysis due to its distributed architecture, configuration complexity, and strong hardware–software dependencies.

The student will investigate how AI models can:

• Identify variability points in robotics codebases and configuration artifacts.
• Assist developers in selecting valid software configurations for specific robot setups.
• Explain architectural dependencies and constraints induced by different configurations.
• Support evolution tasks such as adapting software when hardware components change.

The project is software engineering–driven: the focus is on program analysis, configuration reasoning, architectural understanding, and developer support, rather than on robot control or hardware implementation.

This project contributes:

• An AI-assisted approach for discovering and documenting variability in robotics software systems.
• Empirical evaluation of LLMs for configuration reasoning and architectural explanation tasks.
• A prototype tool that supports developers in configuring and evolving robotics software.
• Insights into the strengths and limitations of LLMs for complex, configuration-heavy software systems.

Robotics knowledge is not required. The project emphasizes software analysis, AI-assisted reasoning, and empirical evaluation using real-world software repositories.

Available spots: 2

Pointers to literature

• Hou, X. et al. Large Language Models for Software Engineering: A Systematic Literature Review. ACM TOSEM, 2024.
• Kästner, C. et al. Variability-Aware Software Analysis. FSE, 2012.
• Timperley, C. et al. ROBUST: A Dataset of Real Bugs in the Robot Operating System. Empirical Software Engineering, 2024.
• Chen, Z. et al. How Well Do Large Language Models Understand Software Architecture? ICSE 2024.