Robot visual sliding mode servoing using SIFT features

To solve the robot visual servoing problem in the natural scene, a robot visual sliding servoing method using SIFT features in the framework of epipolar-geometry-based visual servoing is proposed. Firstly, corresponding image feature points from the current image feature and the desired image are obtained by SIFT algorithm; Then the features error of the whole robot visual servoing system are calculated according to the corresponding points; Finally, considering characteristics of epipolar geometry for robot visual servoing system, the rotation sliding mode controller and translation sliding mode controller are sequentially designed to calculate the rotation movement and translation movement of the robot end-effector. Visual positioning experiments on a 6DOF industrial robot demonstrate that the proposed method can effectively locate an object in the natural scene.