Discrete event simulation of a multi robot pick and place system for reinforcement learning research

Abstract

This poster described a implementation of simulation environment for a Pick and Place conveyer belt for multi robot system based on the Gym library to be used to train Reinforcement Learning agents - see figure below. Our goal is to build this simulation as close as real world conveyer with robots working system. The factors included inside simulation are robot arm speed, conveyor speed, the probability of part distribution, conveyer length , conveyor width, 5 scheduling disciplines for pick and place and robots workspace size.