Engineering Adaptive Fault-Tolerance Mechanisms for Resilient Computing on ROS

Systems are expected to evolve during their service life in order to cope with changes of various natures, ranging from fluctuations in available resources to additional features requested by users. For dependable embedded systems, the challenge is even greater, as evolution must not impair dependability attributes. Resilient computing implies maintaining dependability properties when facing changes. Resilience encompasses several aspects, among which evolvability, i.e., the capacity of a system to evolve during its service life. In this paper, we discuss the evolution of systems with respect to their dependability mechanisms, and show how such mechanisms can evolve accordingly. From a component-based approach that enables to clarify the concepts, the process and the techniques to be used to address resilient computing, in particular regarding the adaptation of fault tolerance (or safety) mechanisms, we show how Adaptive Fault Tolerance (AFT) can be implemented with ROS. Beyond implementation, we draw the lessons learned from this work and discuss the limits of this runtime support to implement such resilient computing features in embedded systems.