A real-time agent model in an asynchronous-object environment

Abstract

To build intelligent control systems for real-life applications, we need to design software agents which combine cognitive abilities to reason about complex situations, and reactive abilities to meet hard deadlines. We propose an operational agent model which mixes AI techniques and real-time performances. Our model is based on an ATN (Augmented Transition Network) to dynamically adapt the agent’s behavior to changes in the environment. Each agent uses a production system and is provided with a synchronization mechanism to avoid the possible inconsistencies of the asynchronous execution of several rule bases. Our agents communicate by message-passing and are implemented in an asynchronous-object environment. We report on the use of our agent model in intensive care patient monitoring.