Fusion of Vision and Inertial Sensors for Position-Based Visual Servoing of a Robot Manipulator

摘要

This paper presents a new technique to combine measurements from both a vision system and an inertial sensor mounted on a robot tip when the robot moves towards an object. The proposed scheme is designed based on a combination of the relative pose (position and orientation) of vision system with information of the inertial sensor (acceleration and angular velocity) to get final relative pose. In addition, a Kalman filter is used to handle asynchronous information from those sensors. The proposed scheme not only can increase the robustness of the pose estimation but also can smooth the velocity control of robot. A computer simulation results are shown to verify the effectiveness of the proposed algorithm.