Abstract
The paper considered the processes of planning and deployment of robot movement by developing an approach to creating a system based on neural networks. A system is proposed that can perceive the environment and controls the movement of the robot by generating correct control commands. For this purpose, 3 tasks were solved, namely, the analysis of the environment in order to determine its features, the determination of the trajectory in order to neutralize the collision, and the determination of controlled influences for the executive bodies in order to implement the movement. The functionality and structure of the neural network for solving each of the tasks is proposed. The proposed approach is compared with existing approaches on key parameters, such as the execution time of the planned movement and the time of calculating the movement trajectory
Publisher
National Academy of Sciences of Ukraine (Co. LTD Ukrinformnauka) (Publications)
Reference8 articles.
1. 1. Siciliano, B. and Sciavicco, L. (2016). Robotics: Modelling, Planning and Control. Springer. http://doi.org/10.1007/978-1-84628-642-1
2. 2. Orr, James & Dutta, Ayan. (2023). Multi-Agent Deep Reinforcement Learning for Multi-Robot Applications: A Survey. https://doi.org/10.3390/s23073625
3. 3. Rivlin, O. (2020, May 5). Generalized Planning With Deep Reinforcement Learning. arXiv.org. https://arxiv.org/abs/2005.02305
4. 4. Niku, S.B. (2010). Industrial Robotics: Programming, Simulation and Applications. John Wiley & Sons. https://doi.org/10.5772/40
5. 5. LaValle, S. M. (2006). Planning algorithms. Cambridge University Press. https://doi.org/10.1017/CBO9780511546877