An Online Fault Injection Framework for Franka Collaborative Robot Based on ROS-Gazebo Simulation

Abstract The market for intelligent industrial robotics is expanding at a faster rate thanks to the prompt development of Industry 5.0. As part of the thesis project at Institute of Industrial Automation Technology and Software Engineering (IAS), the hardware demonstrator including Franka Emika Panda Manipulator and Framos Depth camera D435e, are to be researched. The research scope of this thesis project focuses on several parts, including constructing simulation environment in Gazebo, object detection and robotic arm pick and place movement based on OpenCV by color, real time fault injection (FI) and so on. Precisely speaking, it is needed to build not just the Franka Emika Panda manipulator model but also a special camera model to imitate the function and configuration of real Framos depth camera D435e as accurate as possible. The object detection according to different color can be achieved with functions and algorithms from OpenCV. For pick and place motion, robotic arm will move only after acquiring the coordinates based on the robot base frame which is calculated in combination of camera matrix and hand-eye matrix. The most important content of this thesis project is to design a unique and functional fault injection technique. This section can be divided into 4 main parts which are real time fault injection framework, fault injection graphical user interface (GUI), automatic random fault injection, fault injection tool real camera/robotic arm application respectively. This technique can be implemented both in simulation environment and real scenario. There are two types of fault injection techniques, one is fault injection GUI and the other is automatic random FI. For GUI, it allows users to select a specific fault to inject in a visualized way, while the automatic random FI enables both users and developers to randomly choose a fault and inject random times.