Development in ROS2 of visual servoing control for IDRA

In the space sector, robotics is expected to play an increasingly central role, especially in the field of OSAM, which includes on-orbit assembly, handling, and debris management activities. Space robotics technologies make it possible to limit space debris and replace human intervention in complex or dangerous operations. Furthermore, inflatable technology applied to space is a topic of growing interest, as it enables structures to be launched in a highly compact form and later deployed to full size, significantly reducing both payload volume and launch mass. This thesis was conducted at Politecnico di Torino, in the DIMEAS department, where IDRA, a planar space manipulator prototype, is being developed and tested in simulated microgravity conditions. Air bearings have been installed beneath each motor and connected to a pneumatic network that supplies the necessary airflow. This airflow creates a thin air cushion between the bearings and the floor, effectively limitating friction between the two surfaces. The robot has three joints for three degrees of freedom: two translational and one rotational, which allow full motion within the plane. IDRA adopts a hybrid structure that combines the strengths of both inflactable booms and traditional robotics: the inflatable links provide compactness during launch, while electric motors ensure accuracy and operational func- tionality like rigid robots. This work contributes to the development of a control system for this manipulator, implemented first as a standalone C++ application, and then adapted to the ROS 2 framework. The latter was chosen for its modular architecture, real-time capabilities, and robust integration with hardware via the ros2 control framework and the ability to run simulations in Coppeliasim. A visual servoing control strategy is implemented to detect and track an ArUco marker in real-time. Image processing was implemented using OpenCV to extract the target’s position and orientation from the camera feed. Based on this information, the trajectory of the end-effector was computed by generating a trapezoidal velocity profile and mapping it into joint space, while taking into account the robot’s kinematic and dynamic constraints. The controller manages singularities and improves trajectory tracking performance through careful tuning of PID controllers. During the experimental testing phase, the robotic manipulator demonstrated its ability to detect the target using visual data and autonomously plan and execute a trajectory to reach it. Future work may involve the development and integration of a dedicated control strategy for the end-effector, to enhance the system’s manipulation capabilities and the integration of a camera on the base of the robot.