Using Deep Reinforcement Learning For Robot Arm Control

摘要

Reinforcement learning is a well-proven and powerful algorithm for robotic arm manipulation. There are various applications of this in healthcare, such as instrument assisted surgery and other medical interventions where surgeons cannot find the target successfully. Reinforcement learning is an area of machine learning and artificial intelligence that studies how an agent should take actions in an environment so as to maximize its total expected reward over time. It does this by trying different ways through trial-and-error, hoping to be rewarded for the results it achieves. The focus of this paper is to use a deep reinforcement learning neural network to map the raw pixels from a camera to the robot arm control commands for object manipulation.