In this paper an efficient and low-cost robot grasping system is described. This system efficiently performs grasping tasks of known objects in household environments using just several low-cost devices. For regular visual servo technology that usually applying high-precision but high-cost devices, the target can be localized precisely and the goal arm posture for grasping the target can be calculated accurately, and then the arm can move directly to grasp the target. But for low-cost devices which have very limited performance, it is very difficult to apply this regular technology. In order to overcome this shortage, a new scheme combining machine vision, robotic kinematics and motor control is presented. Using a head camera and an arm camera, the whole grasping problem is split into tow subproblems and then be solved one by one. Our system is tested by performing a representative task in household environments—grasping a coke out of the fridge. The results illustrate that our grasping system accomplishes this kind of grasping task efficiently and successfully.