Simplifying peripheral integration in ROS for manufacturing

This report summarizes the design and implementation of software to reduce the effort spent integrating peripheral devices into robotic workcells. The motivating application is completion of multiple manufacturing tasks undertaken by Los Alamos National Laboratory (LANL) as part of the Plutonium Sustainment Program. Pit manufacturing in LANL is currently completed inside gloveboxes to protect technicians from radioative contamination. This causes ergonomic strain on operators who must move delicate, heavy objects with their arms fully extended. Furthermore, there is an interest in using the current research-oriented manufacturing capabilities to allow for production scale manufacturing while minimizing infrastructure investments. The project undertaken is a demonstration using modern technology to eliminate human handling of radioactive components and creating a quicker and safer automated manufacturing process. To accomplish this, a multi-use manufacturing glovebox was developed and demonstrated using a Yaskawa SIA5 7 Degree-of-Freedom (DoF) industrial manipulator, Robotiq [superscript TM] three finger gripper, five speed drill press, and three-jaw chuck. The goal is to perform multiple manufacturing tasks in a single glovebox to improve throughput while minimizing infrastructure, integration, and logistical complexity. Actions decisions for these devices include data from Force-Torque sensors, pressure sensors, a 3D depth camera, and sensors (encoders, etc.) on the robot. While the peripheral devices are implemented task agnostically, the manufacturing task demonstration utilizing these peripherals include - but are not limited to - drilling, press fitting, and polishing. Aluminum props were used due to their relatively low cost and because fcc [delta] -phase plutonium alloys exhibit some similar mechanical properties to aluminum [14]. To automate the drilling task a Moog Animatics Smartmotor was integrated with the drill press. The 3-jaw chuck was also automated using an Arduino mega 2560 board with a VNH5019 motor driver shield and the ability to measure the current drawn by the DC motor. Object information is relayed by the 3-D depth camera using an object detection algorithm developed at the University of Texas. A finite state machine ensures that operators are unable to perform tasks that would jeopardize plutonium castings and botch subsequent tasks. To achieve the necessary peripheral integration and reduce the burden on future developers responsible for executing related tasks, a software architecture is presented and was developed and written in C++ using the Robot Operating System (ROS) to control the manipulator, the peripherals above, and easily add additional peripherals in the future. A general approach was taken to developing a complete ROS peripheral architecture, and development focused on aspects of the framework suited for industrial devices such as actuators and sensors. Successful implementation into the glovebox manufacturing project validates the proposed software’s potential to minimize the learning curve and standardize implementation of peripherals thus reducing integration time and effort via re-use and extensibility.Mechanical Engineerin