Action Failure Recovery via Model‐based Diagnosis and Conformant Planning

Abstract

A plan carried on in the real world may be affected by a number of unexpected events, plan threats, which cause significant deviations between the intended behavior of the plan executor (i.e., the agent) and the observed one. These deviations are typically considered as action failures.This paper addresses the problem of recovering from action failures caused by a specific class of plan threats: faults in the functionalities of the agent. The problem is approached by exploiting techniques of the Model-Based Diagnosis (MBD) for detecting failures (plan execution monitoring) and for explaining these failures in terms of faulty functionalities (agent diagnosis). The recovery process is modeled as a replanning problem aimed at fixing the faulty components identified by the agent diagnosis. However, since the diagnosis is in general ambiguous (a failure may be explained by alternative faults), the recovery has to deal with such an uncertainty. The paper advocates the adoption of a conformant planner, which guarantees that the recovery plan, if it exists, is executable no matter what the actual cause of the failure is.The paper focuses on a single agent performing its own plan, however the proposed methodology takes also into account that agents are typically situated into a multi-agent scenario and that commitments between agents may exist. The repair strategy is therefore conceived to overcome the causes of a failure while assuring the commitments an agent has agreed with other team members.

Robust plan execution via reconfiguration and replanning

“…To the best of our knowledge, previous approaches to robust plan execution that explicitly deal with resources are mainly off-line ( e.g., [20,34,35,6,5]). On-line methods (e.g., [40,9,14,11,24]), instead, are mostly focused on the achievement of the (propositional) plan goals by interleaving plan execution and (re)planning.…”

“…Another recent work strictly related to the plan repair problem is [24]. This work, however, is mainly focused on the problem of diagnosing the anomalous execution of a plan.…”

Cooperative Monitoring to Diagnose Multiagent Plans

“…We conclude the section by briefly exemplifying the concepts introduced so far. In particular, we present here the office-like domain we used as a test scenario for our experiments (Section 6), this domain is similar to the ones adopted by Micalizio (2013) and Steinbauer and Wotawa (2008). In this domain, robotic agents deliver parcels to the desks of a number of clerks.…”

“…Some recent works (see e.g., Mi & Scacchi, 1993;Gupta, Roos, Witteveen, Price, & de Kleer, 2012;Micalizio, 2013) have argued that a plan repair procedure can be more effective when the causes of the plan failure are known (e.g., identified via diagnosis).…”

“…In this paper we aim at relaxing these assumptions by extending the relational-based framework introduced by Micalizio andTorasso (2008, 2009). Similarly to Roos and Witteveen, we adopt an explicit representation of the MAP at hand in term of agents' actions and shared resources.…”