Vision-based grasping method for 7-DOF manipulator using reinforcement learning
https://doi.org/10.35266/1999-7604-2025-1-5
Abstract
The article describes the solution to the inverse kinematics problem for object grasping with the 7-DOF Franka Emika Panda manipulator, implemented with computer vision. In proposed solution, the robotic arm operates in a physical simulation environment, utilizing reinforcement learning algorithms from machine learning, supplemented by computer vision algorithms for geometric structure-based target localization and
training, enabling the implementation of the entire algorithmic process. The process of problem solving, constructing the corresponding environment, and analyzing the outcomes of the integrated algorithm, which incorporates neural networks, demonstrates its capability to effectively solve complex inverse kinematics tasks. This cost-effective modern technique applies widely to similar tasks with other robotic manipulators.
Keywords
Supporting agencies: the work is supported by the China Scholarship Council (CSC) No. 202108090230
References
1. He Y., Liu S. Analytical inverse kinematics for Franka Emika Panda – A geometrical solver for 7-DOF manipulators with unconventional design // Proceedings of 9th International Conference on Control, Mechatronics and Automation (ICCMA), 2021. IEEE, 2021. p. 194–199.
2. Bullet Real-Time Physics Simulation. URL: https://pybullet.org/wordpress/ (дата обращения: 14.01.2025).
3. Silver D., Hubert T., Schrittwieser J. et al. A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play // Science. 2018. Vol. 362, no. 6419. P. 1140–1144.
4. Torrado R. R., Bontrager P., Togelius J. et al. Deep reinforcement learning for general video game AI // Proceedings of IEEE Conference on Computational Intelligence and Games (CIG), 2018. IEEE, 2018. P. 1–8.
5. Гусев О. В. Решение прямой задачи кинематики для шестизвенного робота-манипулятора // Вестник кибернетики. 2024. Т. 23, № 2. С. 39–48.
6. Li H., Zhao Zh., Lei G. et al. Robot arm control method based on deep reinforcement learning // Journal of System Simulation. 2019. Vol. 31, no. 11. P. 2452–2457.
7. Malik A., Lischuk Y., Henderson T. et al. A deep reinforcement-learning approach for inverse kinematics solution of a high degree of freedom robotic manipulator // Robotics. 2022. Vol. 11, no. 2. P. 44–61.
8. Calderón-Cordova C., Sarango R., Castillo D. et al. A deep reinforcement learning framework for control of robotic manipulators in simulated environments // IEEE Access. 2024. Vol. 12. P. 103133–103161.
9. Franka Emika Panda. URL: https://robodk.com.cn/robot/ru/Franka/Emika-Panda (дата обращения: 14.01.2024).
10. Lillicrap T. P., Hunt J. J., Pritzel A. et al. Continuous control with deep reinforcement learning // arXiv preprint arXiv:1509.02971. 2015.
11. Fujimoto S., Hoof H., Meger D. Addressing function approximation error in actor-critic methods // Proceedings of the 35th International conference on machine learning, 2018, Stockholm. PMLR, 2018. p. 1587–1596.
12. Schulman J., Wolski F., Dhariwal P. et al. Proximal policy optimization algorithms // arXiv preprintarXiv:1707.06347. 2017.
Review
For citations:
Cao Y. Vision-based grasping method for 7-DOF manipulator using reinforcement learning. Proceedings in Cybernetics. 2025;24(1):31-38. (In Russ.) https://doi.org/10.35266/1999-7604-2025-1-5
Views:
159
Email this article 