Intelligent, Fault Tolerant Control for Autonomous Systems

Abstract

We present a methodology for intelligent control of an autonomous and resource constrained embedded system. Geared towards mastering permanent and transient faults by dynamic reconfiguration, our approach uses rules for describing device functionality, valid environmental interactions, and goals the system has to reach. Besides rules, we use functions that characterize a goal's target activity profile. The target activity profile controls the frequency our system uses to reach the corresponding goal. In the paper we discuss a first implementation of the given methodology, and introduce useful extensions. In order to underline the feasibility and effectiveness of the presented control system, we present a case study that has been carried out on a prototype system.