Towards a Natural Language Interface for Flexible Multi-Agent Task Assignment

• DOI: 10.1609/aaaiss.v2i1.27665
• Publication Date: 2024-01-23T00:56:41Z
• Authors: Jake Brawer; Kayleigh Bishop; Bradley Hayes; Alessandro Roncone
• Keywords: Interface (matter); Natural language understanding; Task Analysis; Implementation

Task assignment and scheduling algorithms are powerful tools for autonomously coordinating large teams of robotic or AI agents. However, the decisions these system make often rely on components designed by domain experts, which can be difficult non-technical end-users to understand modify their own ends. In this paper we propose a preliminary design flexible natural language interface task system. The goal our approach is both grant users more control over system's decision process, as well render transparent. Users direct via commands, applied constraints mixed-integer linear program (MILP) using model (LLM). Additionally, proposed alert potential issues with engage them in corrective dialogue order find viable solution. We conclude description planned user-evaluation simulated environment Overcooked describe next steps towards developing transparent allocation