PRIMENA DUBOKOG UČENJA SA PODSTICAJEM ZA PLANIRANJE KRETANJA ROBOTSKOG MANIPULATORA
DOI:
https://doi.org/10.24867/11IH03MilicKeywords:
Deep Reinforcement Learning, motion planning, neural networks, roboticsAbstract
In this paper we present Deep Reinforcement Learning framework as a way to solve problems of motion planning in robotics. With Deep Reinforcement Learning it is possible to generalize motion planning algorithms for both industrial and non-industrial enviroments. In the end, we compared both stochastic and deterministic algorithms for solving motion planning problem.
References
[1] https://www.roboticsbusinessreview.com/manufacturing/deep-learning-factory-automation/ (pristupljeno u martu 2020.)
[2] R. S. Sutton, A. G. Barto “Reinforcement learning, an introduction“, MIT Press, Cambridge, MA, 2018.
[3] T. P. Lillicrap, J. J. Hunt, A. Pritzel, N. Heess, T. Erez, Y. Tassa, D. Silver, D. Wierstra, “Continous Control with Deep Reinforcement Learning”, ICLR, 2016..
[4] T. Haarnoja, A. Zhou, P. Abbeel, S. Levine, “Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor“, ICML, 2018.
[2] R. S. Sutton, A. G. Barto “Reinforcement learning, an introduction“, MIT Press, Cambridge, MA, 2018.
[3] T. P. Lillicrap, J. J. Hunt, A. Pritzel, N. Heess, T. Erez, Y. Tassa, D. Silver, D. Wierstra, “Continous Control with Deep Reinforcement Learning”, ICLR, 2016..
[4] T. Haarnoja, A. Zhou, P. Abbeel, S. Levine, “Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor“, ICML, 2018.
Downloads
Published
2020-12-31
Issue
Section
Mechatronics
