Abstract
In this paper, a reinforcement learning algorithm based on policy search methods has been developed for an anthropomorphic hand-arm system. The advantages of searching in the space of a policy are that the convergence and fastness of the algorithm rely on the proper choice of the policy and of its initial parameters together with the choice of the reward function. The main contribution of the paper consists in the computation of a suitable policy based on the synergies of the hand-arm system, and in the development of a supervised learning strategy to define the initial parameters of the policy. Finally, the reward function has been chosen to evaluate the goodness of the grasp in a synergy-based framework. The experiments executed on the robotic hand-arm system, constituted by the SCHUNK 5-Finger Hand and the KUKA LWR4+, demonstrate the convergence of the method and the promising results obtained for the execution of grasping tasks.
Similar content being viewed by others
References
Ficuciello F, Siciliano B (2016) Learning in robotic manipulation: the role of dimensionality reduction in policy search methods. Comment on Hand synergies: integration of robotics and neuroscience for understanding the control of biological and artificial hands by Marco Santello, Matteo Bianchi, Marco Gabiccini, Emiliano Ricciardi, Gionata Salvietti et al. Physics of Life Reviews
Ficuciello F (2018) Synergy-based Control of Underactuated Anthropomorphic Hands. IEEE Trans Ind Inform (in press)
Suarez R, Rosell J, Garcia N (2015) Using synergies in dual-arm manipulation tasks. In: Proceedings of the 2000 IEEE international conference on robotics and automation
Billard A, Calinon S, Dillmann R, Schaal S (2008) Robot programming by demonstration. In: Siciliano B, Khatib O (eds) Springer handbook of robotic. Springer, Berlin
Argall BD, Chernova S, Veloso M, Browning B (2009) A survey of robot learning from demonstration. Robot Auton Syst 57(5):469–483
Kormushev P, Calinon S, Caldwell D (2007) Reinforcement learning in robotics: applications and real-world challenges. Robotics 2:122–148
Romero J, Feix T, Ek C, Kjellstrom H, Kragic D (2013) A metric for comparing the anthropomorphic motion capability of artificial hands. IEEE Trans Robot 29:1342–1352
Theodorou E, Buchli J, Schaal S (2010) Learning policy improvements with path integrals. J Mach Learn Res 9:828–835
Stulp F, Theodorou A, Schaal S (2012) Reinforcement learning with sequences of motion primitives for robust manipulation. IEEE Trans Robot 28(6):1360–1370
Hansen N (2006) The CMA evolution strategy: a comparing review. In: Lozano J, Larranaga P, Inza I, Bengoetxea E (eds) Towards a new evolutionary computation. Springer, Berlin, pp 75–102
Stulp F, Sigaud O (2012) Path integral policy improvement with covariance matrix adaptation, In: Proceedings of the 29th international conference on machine learning, Edinburgh, Scotland, pp 2112–2118
Theodorou E, Buchli J, Schaal F (2010) A generalized path integral control approach to reinforcement learning. J Mach Learn Res 11:3137–3181
Kormushev P, Calinon S, Caldwell D (2010) Robot motor skill coordination with EM-based reinforcement learning, In: Proceedings IEEE/RSJ international conference on intelligent robotics systems, Taipei, Taiwan, pp 3232–3237
Rombokas E, Malhotra M, Theodorou E, Todorov E, Matsuoka Y (2012) Reinforcement learning and synergistic control of the act hand. IEEE/ASME Trans Mechatron 18(2):569–577
Santello M, Flanders M, Soechting J (1998) Postural hand synergies for tool use. J Neurosci 18(23):10105–10115
Bicchi A (1994) On the closure properties of robotic grasping. Int j Robot Res 14(4):319–334
SCHUNK Hand webpage. http://mobile.schunk-microsite.com/en/produkte/produkte/servoelektrische-5-finger-greifhand-svh.html
Ficuciello F, Zaccara D, Siciliano B (2016) Synergy-based policy improvement with path integrals for anthropomorphic hands. In: IEEE/RSJ international conference on intelligent robots and systems, Daejeon, Korea, 9–14 September
Ficuciello F, Palli G, Melchiorri C, Siciliano B (2014) Postural synergies of the ub hand iv for human-like grasping. Robot Auton Syst 62:357–362
Palli G, Melchiorri C, Vassura G, Scarcia U, Moriello L, Berselli G, Cavallo A, Maria GD, Natale C, Pirozzi S, May C, Ficuciello F, Siciliano B (2014) The dexmart hand: mechatronic design and experimental evaluation of synergy-based control for human-like grasping. Int J Robot Res 33:799–824
Jarrasse N, Ribeiro AT, Sahbani A, Bachta W, Roby-Brami A (2014) Analysis of hand synergies in healthy subjects during bimanual manipulation of various objects. J Neuroeng Rehabil 11:113
Ficuciello F, Zaccara D, Siciliano B (2016) Learning grasps in a synergy-based framework. In: 15th International symposium on experimental robotics, Tokyo, Japan, 3–6 October
schunk\_svh\_driver. http://wiki.ros.org/schunk_svh_driver
Deisenroth M, Rasmussen CE (2011) Pilco: a model-based and dataefficient approach to policy search. In: Proceedings of the 28th international conference on machine learning, pp 465472
Saveriano M, Yin Y, Falco P, Lee D (2017) Data-efficient control policy search using residual dynamics learning. In: IEEE/RSJ international conference on intelligent robots and systems, Vancouver, BC, 2017, pp. 4709–4715
Acknowledgements
The author wants to thank Damiano Zaccara which contributed to the development of the set-up and supported the experiments described in this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
A preliminary version of this work has been published in IROS 2016 Proceedings. The research leading to these results has been partially supported by the RoDyMan project, funded by the European Community’s Seventh Framework Programme (FP7/2007-2013) under ERC AdG-320992, and partially by MUSHA project, National Italian Grant under Programma STAR Linea 1.
Rights and permissions
About this article
Cite this article
Ficuciello, F. Hand-arm autonomous grasping: synergistic motions to enhance the learning process. Intel Serv Robotics 12, 17–25 (2019). https://doi.org/10.1007/s11370-018-0262-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11370-018-0262-0