Dealing with Failures for Execution Consistency in Context-aware Systems.

In this paper, we consider a symbolic mechanism designed to help guide the performance of actions of a Belief-Desire-Intention (BDI) agent under both concurrency and relevance criteria. Behind the planning activity which can optimistically estimate the relevance of traces, we propose to directly supervise the best effort execution of the intentions as to keep track of the most relevant ones in case of failures. To support the presented method, a typical use case is given that targets the automated guidance of an autonomous vehicle. [ABSTRACT FROM AUTHOR]