Optimization of the Research of an Unknown Object in an Unknown Environment

AbstractThis article addresses the problem of the research of an unknown object in an unknown environment using visual servoing. The system is camera-in-hand system mounted directly on the robot manipulator, and a camera-fired system fixed on the top of the mobile platform and observes the scene from there. The method proposed is based on the epipolar constraint defined between the two cameras. While eye-in-hand camera covers the epipolar line, we compute the next best view to be observed. The system executes the task between two positions blindly, using a velocity control law which generates a safety movement for the manipulator robot. The proposed approach minimizes the computational time of the training and takes into account the safety and security of the environment. The paper first gives an overview of the existing approach for localizing an unknown object and evaluates some of those approaches and propose a novel approach