Blackboard Architecture for Detecting and Notifying Failures for Component-Based Unmanned Systems

Abstract

This paper describes the blackboard architecture that is capable of detecting the component failures/recoveries in the component-based unmanned systems and notifying them to the associated components. The blackboard architecture monitors each component of the system in order to detect its failures/recoveries at runtime and identify the causes of failures. Using the dependency relationships between components, the blackboard architecture performs impact analysis so that it determines the scope of failure/recovery notification in the components of the system. The notification messages delivered to the components can trigger safety actions against the failures. The prototypes of blackboard architecture have been developed for Microsoft Robotics Developer Studio (MSRDS) based unmanned systems and Robot Operating System (ROS) based unmanned systems. The prototypes are used to validate the blackboard architecture with an unmanned ground vehicle (UGV) system and a patrolling robot system as case studies.