Alignment Method of Combined Perception for Peg-in-Hole Assembly with Deep Reinforcement Learning

Algorithm 1

System pipeline.

1: Initialize replay buff

2: Initialize evaluation network parameters

3: Initialize target network parameters

4: for episode=1, do

5:  for, do

6:   Obtain image from environment

7:   With probability select a random adjustment action

8:   otherwise select adjustment action

9:   Execute adjustment action in CoppeliaSim

10:   Obtain image and reward from environment

11:   Store transition in

12:   Sample random minibatch of transitions from

13:   Set

14:   Perform a gradient descent step on

15:  end for

16: end for