Implicit Communication in Human-Machine Collaboration
Abstract
The goal of this project was to advance human-machine collaboration, specifically focusing on the machine's ability to understand and partake in implicit communication. We focused much of the work on inferring underlying human objectives and preferences, from learning from different types of input (physical corrections, feature traces, even one single state), to introducing better models of human behavior that lead to better inference, to preventing wrong/harmful inference. We also made a number of contributions on enabling robots to communicate themselves implicitly: objectives, capability/incapability, their future task plans, and even emotional state. We have published numerous conference and journal papers, and received 4 best paper nominations (three times at HRI, the premier conference inhuman-robot interaction, and once at TRO).
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 10, 2022
- Accession Number
- AD1190026
Entities
People
- Anca Dragan
Organizations
- University of California Regents