Back to Reality for Imitation Learning

Abstract

Imitation learning, and robot learning in general, emerged due to breakthroughs in machine learning, rather than breakthroughs in robotics. As such, evaluation metrics for robot learning are deeply rooted in those for machine learning, and focus primarily on data efficiency. We believe that a better metric for real-world robot learning is time efficiency, which better models the true cost to humans. This is a call to arms to the robot learning community to develop our own evaluation metrics, tailored towards the long-term goals of real-world robotics.

Published at CoRL 2021, blue sky oral track

References in corpus (7)

• Learning from Suboptimal Demonstration via Self-Supervised Reward Regression
• Visual Imitation Made Easy
• Learning Dexterous Manipulation from Suboptimal Experts
• Modeling Long-horizon Tasks as Sequential Interaction Landscapes
• ACNMP: Skill Transfer and Task Extrapolation through Learning from Demonstration and Reinforcement Learning via Representation Sharing
• Generalization Guarantees for Imitation Learning
• Learning Multi-Stage Tasks with One Demonstration via Self-Replay