Robotic Arm Representation Using Image-Based Feedback for Deep Reinforcement Learning
1. 摘要
This paper presents a technique that uses feedback images for implementing a robotic task using Deep Reinforcement Learning (DRL). The agent (controller) type is Actor-Critic that uses Temporal- Difference (TD) to gain knowledge about the environment (plant). The representation of the robot is implemented in the V-Rep simulator.
本文提出了一种使用反馈图像通过深度强化学习（DRL）来实现机器人任务的技术。 Agent（控制器）类型是Actor-Critic，它使用时间差异（TD）来获取有关环境（plant）的知识。 机器人的表示在V-Rep仿真器中实现。