Sous Chef? Quickly Teaching Food Preparation Tasks to an Autonomous Robot Team

Autonomous robots have reached the physical capabilities to perform useful tasks in mixed-initiative human-robot teams. In this paper we provide an overview of the latest version of our DIARC architecture and show how it can be used in practical setting to quickly instruct a group of robots to perform novel tasks, in this instance food preparation tasks. Specifically, we motivate essential architectural capabilities required for effective mixed-initiative human-machine teaming and briefly discuss how DIARC meets them. We then introduce as an example of practical application a multi-robot, multi-human food assembly task as it occurs in fast food restaurants and show how different aspects of the task can be quickly taught through natural language dialogues, with the system immediately able to perform those tasks.