Distributed (Re)Planning With Preference Information
Pieter Buzing, Cees Witteveen · 2005
Many planning problems have a distributed nature: different parties have to co-operate in order to make efficient use of shared resources, while not violating any constraints. Additionally, there often are dynamic aspects to be considered. Due to e.g., unexpected changes in the (outside) world or changes of individual goals of the agents each of these actors might come up with a corresponding change of their constraints. As a result, the problem might require some form of replanning of the agent activities such that all constraints can be satisfied. We discuss a method to deal with changes of temporal constraints in a distributed Simple Temporal Network using preference information. We adapt an existing method to obtain a consistent plan using distributed temporal constraints such that all preferences of all agents above a certain level are satisfied, and changes to the constraints or preferences can be dynamically incorporated, adapting the (inconsistent) plan in such a way that all constraints of some preference level are satisfied. 1