Simulation Environment for Modeling and Testing of Autonomous Assembly in Space for Multiple Robotic Arms

Multi-arm robotic assembly in space is an increasingly popular research topic as robotic arms can be used to decrease repair costs. We propose a simulation framework to model different scenarios of multi-arm robotic assembly in space with an ever-changing harsh environment. The end goal of this project is to develop a framework that will be used to expedite development time and reduce costs for deploying and testing autonomous in-orbit satellite repairs. Simulation is a valuable, viable, and low-cost option for testing the developed algorithms without the need to recreate the orbital conditions in an elaborate and expensive laboratory environment. The proposed simulation framework uses the open-source tools: ROS Kinetic and Gazebo 7. We propose a system where robotic arms coordinate and collaborate to perform tasks autonomously, in our case, to drive a screw into a plate. A move-able robot with a 3D sensing camera maps the environment and, using image recognition and localization, notifies the arms of obstacles and the locations of the screw and plate models. The arms calculate a collision-free path to the model, grasp the object, and return to a shared location. A force-torque sensor is used for the autonomous driving of the screw into the plate - completing the assembly task. In addition to the proposed framework, we present two experimental setups accomplished with the framework, an assembly task demonstration and an object recognition, localization, and collision mapping demonstration. The contribution of this work is in the implementation and integration of a wide variety of algorithms and tools into a unified toolset to allow proof-of-concept development targeting robotic assembly in space. The proposed framework is a valuable tool for education or further research and is currently under development at California State University Long Beach.