This paper proposes a visual servoing algorithm for mobile robot navigation based on object profiles which is alternative to approaches based on point correspondences. Motivation to this research is that for unstructured scenes the task of solving correspondences is definitely a harder task than contour detection. An unstructured 3D scene exhibits only surfaces whose main 2D features are their apparent contours projected onto the image plane. Apparent contours are used to estimate the epipolar geometry and some special symmetry conditions of epipoles whereby the visual servoing is able to steer the mobile robot to the desired position.