Task-space sensory feedback information such as visual feedback is used in many modern robot control systems as it improves robustness to model uncertainty. However, existing sensory feedback control schemes are only valid locally in a finite task space within a limited sensing zone where singularity of the Jacobian matrix is avoided. In this paper, the global stability problem of task-space sensory feedback control system is formulated and solved. The proposed method is based on multiple regional feedback information where each feedback information is employed in a local region. The combination of the local feedback covers the entire workspace and thus guarantees the global movement of the robot. In addition, the switching from one feedback information to another is embedded in the controller without using any hard or discontinuous switching. Experimental results are presented to illustrate the performance of the proposed controller.