On-line monitoring and diagnosis of multi-agent systems: a model based approach

Abstract

The paper presents an approach for the monitoring and diagnosis of multi-agent systems where mobile robotic agents provide services and partial observability of the environment is achieved via a set of fixed sensors. This kind of systems exhibits complex dynamics where weakly predictable interactions among agents may arise. A model-based approach to on-line monitoring and diagnosis is adopted: while the dynamics of the system components and their relations are modeled via communicating automata, the global system model is factored in a number of subsystems dynamically aggregating a convenient set of component models. The On-line Monitoring Module (OMM) estimates the possible evolutions of the system by exploiting partial system observability provided by sensors and agents messages and enforces global constraints. When the monitor detects failures in the actions execution, the Diagnostic Interpretation Module (DIM) is triggered for explaining the failure in terms of faults in the robotic agents and/or troublesome interactions among them. As a specific case-study we refer to the RoboCare project2.