Simple Kinematic Calibration Approach for Eye-In-Hand Depth-Camera

This paper presents a simple and universal approach to calibrate all RGB-D based vision system mounted in eye-in hand configuration on a robotic arm. In literature, many approaches have been proposed for the calibration procedure of general vision systems. These methods are designed to estimate extrinsic and intrinsic parameters of the camera. They are comprehensive, but their robust ness comes with a huge computational effort required. The proposed calibration algorithm has been conceived specifically for an RGB-D sensor, a system combining a standard RGB camera and a depth camera. The algorithm makes use of a calibration target with known pose to evaluate the mounting parameters which define the pose of the vision system with respect to the link of the robot on which it mounted. The proposed procedure has been developed for a custom mobile manipulator conceived for assistive indoor applications, but it can be applied in other cases too. The algorithm has been tested and the results, obtained with the calibration procedure, has been proved to be comparable with the acquisitions made using just the RGB-D system.