本論文已被瀏覽 244 次， [ ] 37 次，[ ] [ ]
In this project we propose a framework to model the behavior and evaluate the reliability and safety of robotic surgery systems. A software simulator and associated user interface tools are used to generate the simulated hardware/software data of a robotic system performing interventions on small animals. The main contributions of this work are the usage of Dynamic Bayesian Networks (DBN) to model both software and hardware dynamics of the robotic surgery system, as well as detecting adverse and anomalous (A&A) events in the system. We show empirically that the model can accurately capture aspects of the software/hardware dynamics. Furthermore, we show that the models are able to accurately classify, detect and predict certain kinds of A&A events. Finally, comparison between different models demonstrates the usefulness of modeling both hardware and software state, as opposed to using only hardware state alone.