Asking Easy Questions: A User-Friendly Approach to Active Reward Learning
arXiv:1910.04365
Abstract
Robots can learn the right reward function by querying a human expert. Existing approaches attempt to choose questions where the robot is most uncertain about the human's response; however, they do not consider how easy it will be for the human to answer! In this paper we explore an information gain formulation for optimally selecting questions that naturally account for the human's ability to answer. Our approach identifies questions that optimize the trade-off between robot and human uncertainty, and determines when these questions become redundant or costly. Simulations and a user study show our method not only produces easy questions, but also ultimately results in faster reward learning.
arXiv commentProceedings of the 3rd Conference on Robot Learning (CoRL), October 2019
References in corpus (4)
- Bayesian Active Learning for Classification and Preference Learning
- Near-Optimal Bayesian Active Learning with Noisy Observations
- An Active Learning Algorithm for Ranking from Pairwise Preferences with an Almost Optimal Query Complexity
- The Green Choice: Learning and Influencing Human Decisions on Shared Roads
Cited by in corpus (5)
- The Green Choice: Learning and Influencing Human Decisions on Shared Roads
- Learning Submodular Objectives for Team Environmental Monitoring
- Explaining Reward Functions to Humans for Better Human-Robot Collaboration
- LazyDAgger: Reducing Context Switching in Interactive Imitation Learning
- ROIAL: Region of Interest Active Learning for Characterizing Exoskeleton Gait Preference Landscapes