A SOFTWARE FRAMEWORK FOR ROS BASED ROBOT CONTROL

Abstract

Robotics education relies heavily on the ROS platform. However, due to open source nature of ROS, installation of robotic packages and libraries require knowledge of dependencies and version control. A software framework collecting and integrating all of the required packages can facilitate learning. An educational framework was built by compiling and virtualizing robotic packages to improve the efficiency of learning robotics for a novice. A review of the frameworks demonstrates that, although Virtual Machine virtualization and Docker containerization both have advantages, a shift towards containerization can be noticed. Three methods for building the framework, VM virtualization, complete framework containerization, and modular process containerization, were tested. Results show that Docker is more resource-efficient with 14-15% average CPU utilization for longer tasks while VM rises up to 30%. Docker’s features make it a great, versatile tool for the framework.