Human hand manipulations by their fingers are carried out by the sensory-motor coordination, as the visual or tactile sensing, and the motor control signals. Although the visual information is relevant for manipulation, there are tasks that humans can even if the object is occluded. Thus, the skin information, muscles, and tendons variations are fundamentals to estimate spatial dimensions of the object to manipulate it. In this paper is exploiting the hemispherical soft tip of the finger to get skillful manipulation of a rigid circular object under the assumption that the measurement of the object angle is not available. The proposed regulator is used to control the contact forces and orientation of the object. Our approach is based in to create a virtual reference frame, fixed at the origin of the world reference frame, to estimate object angle. Stability analysis, based on passivity considerations, is presented without force sensing, object angle measure, not either any knowledge of the deformation nor penetration. Experimental results show the closed-loop performance for the rolling a wheel onto the plane by robotic finger.