Virtual Experimental Investigation for Industrial Robotics in Gazebo Environment

Abstract

Measuring the agility performance of the industrial robots as they are performing in unstructured and dynamic environments is a thought-provoking research topic. This paper investigates the development of industrial robotic simulation algorithms for the effective application of robots in those changing environments. The distributed framework for this investigation is the Robot Operating System (ROS) which is extensively used in robotic applications. ROS-Industrial (ROS I), which extends the capabilities of ROS to manufacturing, allows us to interoperate between industrial robots, sensors, communication buses and other kinds of automation tools. Gazebo is used as the open-source 3D simulator to design a virtual industrial robotic system, which is a prevailing tool as a node in the ROS environment. An effort is underway to replicate the in-house experimental robotic kitting lab with a graphical physics simulation that can be shared worldwide. This graphical physics simulation is not tied to a specific robotic control system. An experimental approach will be presented detailing the issues related to a physics based simulation of kitting with multiple collaborative robots, multiple tools, parts, tool changers, safety system, and sensors. In this realm, the ability for the simulation environment to encompass the current system as well as additional more complex sensors and actuators will be discussed. To make this simulation environment more realistic, Gaussian noise will be introduced to the data generated by virtual sensors. We expect that this experimental approach will be a seamless way for users to verify and validate their control systems even if they do not have a physical robot at their facilities.