Electrical Engineering and Computer Science Faculty Publications

Document Type

Conference Proceeding

Publication Title

International Conference on Agents and Artificial Intelligence

Abstract

Automated planning tools play a large and expanding role in the function of many parts of our lives. The complex nature of the planning problem and the increasing amount of information the human planner must synthesize indicate that assistive automation must soon become the norm. Despite this, many existing automated planners are incapable of producing plans that reflect the desires and expertise of their operators. They do not have the direct ability to consider the operators’ priorities, nor can they exploit expert operational knowledge that comes from human experience and not data systems. In this paper we present methods to learn operator planning preferences and then nudge our automated logistic planner to produce plans that are better aligned with operator preferences without changing the code of the planner. © 2024 by SCITEPRESS - Science and Technology Publications, Lda.

First Page

958

DOI

10.5220/0012423600003636

Publication Date

2024

Download

Share

COinS