Tell me when and why to do it!: run-time planner model updates via natural language instruction

Authors:
Rehj Cantrell;Kartik Talamadupula;Paul Schermerhorn;J. Benton;Subbarao Kambhampati;Matthias Scheutz
Affiliations:
Indiana University, Bloomington, IN, USA;Arizona State University, Tempe, AZ, USA;Indiana University, Bloomington, IN, USA;Arizona State University, Tempe, AZ, USA;Arizona State University, Tempe, AZ, USA;Tufts University, Boston, MA, USA
Venue:
HRI '12 Proceedings of the seventh annual ACM/IEEE international conference on Human-Robot Interaction
Year:
2012

Citing 7
Cited 1

Casper: Space Exploration through Continuous Planning

IEEE Intelligent Systems
Interactive robot task training through dialog and demonstration

Proceedings of the ACM/IEEE international conference on Human-robot interaction
First steps toward natural human-like HRI

Autonomous Robots
Sapa: a multi-objective metric temporal planner

Journal of Artificial Intelligence Research
Robust spoken instruction understanding for HRI

Proceedings of the 5th ACM/IEEE international conference on Human-robot interaction
Planning for human-robot teaming in open worlds

ACM Transactions on Intelligent Systems and Technology (TIST)
The LAMA planner: guiding cost-based anytime planning with landmarks

Journal of Artificial Intelligence Research

Handling open knowledge for service robots

IJCAI'13 Proceedings of the Twenty-Third international joint conference on Artificial Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

Robots are currently being used in and developed for critical HRI applications such as search and rescue. In these scenarios, humans operating under changeable and high-stress conditions must communicate effectively with autonomous agents, necessitating that such agents be able to respond quickly and effectively to rapidly-changing conditions and expectations. We demonstrate a robot planner that is able to utilize new information, specifically information originating in spoken input produced by human operators. We show that the robot is able to learn the pre- and postconditions of previously-unknown action sequences from natural language constructions, and immediately update (1) its knowledge of the current state of the environment, and (2) its underlying world model, in order to produce new and updated plans that are consistent with this new information. While we demonstrate in detail the robot's successful operation with a specific example, we also discuss the dialogue module's inherent scalability, and investigate how well the robot is able to respond to natural language commands from untrained users.