The flexibility of computer vision is attractive when designing manipulation systems which must interact with otherwise unsensed objects. However, occlusions introduce significant challenges to the construction of practical vision-based control systems. This paper provides empirical validation of a vision based control strategy that affords guaranteed convergence to a visible goal from essentially any "safe" initial position while maintaining full view of all the feature points along the way. The method applies to first (quasi-static, or "kinematic") and second (Lagrangian or "mechanical") order plants that incorporate an independent actuator for each degree of freedom.