Human Teachers’ Intentions for Natural Interactive Task Learning

Goal

Propose robot interaction design that allows people to use their natural teaching process to teach robots

Task of Focus

Teach a robot how to build a wall using blocks in a simulated robot environment

Methods

Wizard of Oz Study, Semi-structured interviews

Analysis Methods

Cognitive Task Analysis

Population

10 participants recruited through snowball sampling

Results

• I created a taxonomy of different parts of a teaching instruction that included intention, modality, concept expression type, and domain ontology.
• We proposed robot interaction design framework to support different teaching styles based on SharedPlans, an existing theory of interaction.

Paper

Preeti Ramaraj, Charles L. Ortiz, Jr., and Shiwali Mohan. "Unpacking Human Teachers' Intentions For Natural Interactive Task Learning." In International Symposium on Robot and Human Interactive Communication (RO-MAN 2021). 2021.