Reasoning about Success and Failure in Intentional Agents

Rational agents must be aware of their success and failure to truly assess their own progress towards their intended goals. In this study we describe a detailed investigation of how current BDI agents monitor their successes and failures during their reasoning cycle. Our analysis indicates that the existing architectures are inadequate to specifically detect failures in their own behaviors. This makes them unaware of the reality of the environment in which they are operating. We propose an extended BDI-like architecture to address these problems. We extend the current reasoning cycle by reformulating the execution of actions and plans, and introducing additional rules to detect failures. The resulting reformulation can be applied to existing systems such as JACK, JAM, etc. As a case study we extended JASON to implement the extended BDI architecture.