Interactive Robot Task Learning: Human Teaching Proficiency with Different Feedback Approaches

The deployment of versatile robot systems in diverse environments requires intuitive approaches for humans to flexibly teach them new skills. In our present work, we investigate different user feedback types a real movement skill. We compare as star ratings on an absolute scale single roll-outs versus preference-based pairwise comparisons with respective optimization algorithms (i.e., variation co-variance matrix adaptation -evolution strategy (CMA-ES) and random optimization) the game skill cup-and-ball. experimental investigation users, investigated influence type experience interacting interfaces performance learning systems. While there is no significant difference subjective between conditions, performance. system learned task quicker, but this did not users’ evaluation it. follow-up study, confirmed that indeed can be attributed human