Software Reconfiguration in Robotics (Summary)

Paper i proceeding, 2025

We present our article on our empirical study of software reconfiguration in robotics, which was published as a journal-first article in Empirical Software Engineering (EMSE) [Pe24b]. In this article, we assessed and compared the state of the art and practice in automated software reconfiguration in robotic systems at runtime. To operate in diverse environments, robotic systems need to reconfigure their software, e.g., by loading/unloading software modules or changing their parameters, to cope with changing missions and conditions. We identified a significant mismatch between how reconfiguration is considered in the state-of-the-art, derived from a review of 78 academic publications on reconfiguration in robotics, and the state-of-practice, derived from a review of how reconfiguration is supported by four popular robotics frameworks and what is implemented in 48 robotics (sub)systems on GitHub. While research focuses on complex structural reconfiguration, considering advanced triggers and constraints, in practice only parameter reconfiguration has been widely adopted. We provide practitioners with best practices for developing reconfigurable robotic systems, and researchers with the insights needed to focus on reconfiguration techniques that are applicable in practice.