Quantifying teaching behavior in robot learning from demonstration
Tóm tắt
Learning from demonstration allows for rapid deployment of robot manipulators to a great many tasks, by relying on a person showing the robot what to do rather than programming it. While this approach provides many opportunities, measuring, evaluating, and improving the person’s teaching ability has remained largely unexplored in robot manipulation research. To this end, a model for learning from demonstration is presented here that incorporates the teacher’s understanding of, and influence on, the learner. The proposed model is used to clarify the teacher’s objectives during learning from demonstration, providing new views on how teaching failures and efficiency can be defined. The benefit of this approach is shown in two experiments ([Formula: see text] and [Formula: see text], respectively), which highlight the difficulty teachers have in providing effective demonstrations, and show how [Formula: see text]–180% improvement in teaching efficiency can be achieved through evaluation and feedback shaped by the proposed framework, relative to unguided teaching.
Từ khóa
Tài liệu tham khảo
Cakmak M, 2011, Proceedings of the ICML Workshop on New Developments in Imitation Learning
Khan F, 2011, Neural Information Processing Systems, 24, 1449
Maeda G, 2017, Conference on Robot Learning (CoRL)
Ng AY, 2000, Proceedings of the Seventeenth International Conference on Machine Learning (ICML), 663
Niekum S (2012) ar_track_alvar. http://wiki.ros.org/ar_track_alvar.
Schaal S, 1997, Neural Information Processing Systems, 9, 1040
Zhu X, 2015, Proceedings of the Twenty-Ninth Conference on Artificial Intelligence (AAAI)