Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance
  • View Times: 14
  • |
  • Release Date: 2022-11-26
  • Q-learning
  • DQN
  • SARSA
  • DDQN
  • homogeneous transformation
Video Introduction

This video is adapted from 10.3390/electronics11213636

The video describes the real time experimentation of vision based path planning using TAL BRABO 5 DOF robotic manipulator. The camera in eye to hand configuration detects the start, goal and obstacles placed in robotic workspace and converts the pixels into grid coordinates which is taken as input for the Qlearning, SARSA,DQN and Double DQN reinforcement learning algorithms. The better performance was given by Double DQN so the optimal path obtained were again converted into robot coordinates using coordinate transformation technique. Finally robot traces the learned path smoothly and efficiently. 

Full Transcript
1000/1000

Confirm

Are you sure to Delete?
Cite
If you have any further questions, please contact Encyclopedia Editorial Office.
Singh, A.;  Venkatesan, K.;  Shakeel, M.; Ram, K. Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance. Encyclopedia. Available online: https://encyclopedia.pub/video/video_detail/506 (accessed on 15 April 2024).
Singh A,  Venkatesan K,  Shakeel M, Ram K. Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance. Encyclopedia. Available at: https://encyclopedia.pub/video/video_detail/506. Accessed April 15, 2024.
Singh, Abhilasha, Kalaichelvi Venkatesan, Mohamed Shakeel, Karthikeyan Ram. "Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance" Encyclopedia, https://encyclopedia.pub/video/video_detail/506 (accessed April 15, 2024).
Singh, A.,  Venkatesan, K.,  Shakeel, M., & Ram, K. (2022, November 26). Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance. In Encyclopedia. https://encyclopedia.pub/video/video_detail/506
Singh, Abhilasha, et al. "Vision-Based Bio-Inspired RL for Manipulator Obstacle Avoidance." Encyclopedia. Web. 26 November, 2022.