Agenda

Speaker: Raunak Bag, DAIS, Universitå Ca' Foscari

Abstract:
(PhD defense dry run)
The thesis aims to contribute towards the state-of-the-art in the non-functional requirements (NFRs) elicitation and conflict consideration process for robotic systems. Initially, the research introduces and validates a novel framework, SCARS, designed as a Domain Specific Language (DSL) that extends ROS2 DSL capabilities. SCARS facilitates the analysis of NFR impacts, computing the optimal NFR satisfaction parameters tailored to specific system constraints, as tested in the Gazebo simulation with iRobot® Create®3.
Building upon this, the study introduces a simulation-based methodology for pre-emptive identification of NFR conflicts in different operational contexts. Based on a single-robot system, the simulation results are helpful in inferring and evaluating the different conflicts between NFRs and studying the impact of different contexts on the requirements themselves.
The methodology is further extended to elicit correlations between NFRs and operational environments in multi-robot systems, highlighting how environmental dependencies can lead to unexpected system behaviours post-deployment. Through simulation, this work identifies NFR correlations, providing system designers with insights to foresee and counteract possible threats in uncontrolled environments for multifleet systems.
Lastly, the research extends into the domain of physical robots by proposing enhancements to behaviour trees, focusing on integrating non-functional and operational constraints within the functional goals. Using an interview-based approach, this part of the study profiles the inclusion of NFR considerations in LLM-supported behaviour trees for the RoboCup@Home 2024 challenge. A subsequent methodology proposes LLM-assisted NFR profiling, reducing the dependency on an NFR analyst for profile creation.
Collectively, these studies contribute to a comprehensive understanding of NFR management in robotics, offering methodologies that enhance system reliability, safety, and operational awareness across varied and dynamic environments.