Conclusion
We proposed a TATD-HER curriculum learning method to learn manipulation skills on multi-step complex tasks. The method addresses the complicated task with both explicit and implicit curriculum learning mechanisms. The experimental results demonstrate the effectiveness of our proposed TATD-HER method and show that the combination of explicit and implicit curriculum learning techniques is crucial for learning complex manipulation skills. Future work involves applying our method to more complex tasks.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Liu N-J, Lu T, Cai Y-H, et al. A review of robot manipulation skills learning methods. Acta Autom Sin, 2019, 45: 458–470
Popov I, Heess N, Lillicrap T, et al. Data-efficient deep reinforcement learning for dexterous manipulation. 2017. ArXiv: 1704.03073
Nair A, McGrew B, Andrychowicz M, et al. Overcoming exploration in reinforcement learning with demonstrations. In: Proceedings of IEEE International Conference on Robotics and Automation, Brisbane, 2018. 6292–6299
Bengio Y, Louradour J, Collobert R, et al. Curriculum learning. In: Proceedings of International Conference on Machine Learning, Montreal, 2009. 41–48
Fournier P, Sigaud O, Chetouani M, et al. Accuracy-based curriculum learning in deep reinforcement learning. 2018. ArXiv: 1806.09614
Andrychowicz M, Wolski F, Ray A, et al. Hindsight experience replay. In: Proceedings of Advances in Neural Information Processing Systems, Long Beach, 2017. 5048–5058
Lillicrap T, Hunt J, Pritzel A, et al. Continuous control with deep reinforcement learning. In: Proceedings of International Conference on Learning Representations, 2016
Schulman J, Wolski F, Dhariwal P, et al. Proximal policy optimization algorithms. 2017. ArXiv: 1707.06347
Acknowledgements
This work was supported by National Natural Science Foundation of China (Grant Nos. 61773378, U1713222, U1806204) and Equipment Pre-research Field Fund (Grant No. 61403120407).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Liu, N., Lu, T., Cai, Y. et al. Manipulation skill learning on multi-step complex task based on explicit and implicit curriculum learning. Sci. China Inf. Sci. 65, 114201 (2022). https://doi.org/10.1007/s11432-019-2648-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11432-019-2648-7