The paper “ROSpec: A Domain-Specific Language for ROS-Based Robot Software”, co-authored by Paulo Canelas and Alcides Fonseca, was published in top 10% journal Proceedings of the ACM on Programming Languages (PACMPL) and will be presented at OOPSLA 2025, taking place from October 12 to 18, 2025, in Singapore. This research was led by researchers at LASIGE and Carnegie Mellon University.

The research addresses the challenges of configuring and integrating components in robot software frameworks such as the Robot Operating System (ROS). While these frameworks allow developers to quickly build systems from reusable, off-the-shelf components, they often lack documentation on how components should be configured and integrated. Even when documentation is available, its natural language descriptions are not enforceable, leading to misconfigurations that can cause unpredictable or even dangerous robot behaviours. To address this issue, the authors introduce ROSpec, a ROS-tailored domain-specific language for specifying and verifying component configurations and their integration. ROSpec builds on domain concepts from ROS and is informed by an empirical study of real-world misconfigurations. At its core, ROSpec verifies the correctness of component arguments and configurations, checks communication properties to ensure proper integration, and validates that system configurations respect the assumptions and constraints of their deployment contexts. The authors demonstrate ROSpec’s effectiveness by modelling a medium-sized warehouse robot with 19 components and by manually analysing, categorising, and partially specifying 182 misconfiguration cases gathered from a robotics Q&A platform.

The paper is available here.