We are developing feedback control methods based on the use of Proper Orthogonal Decomposition POD and Modified Linear Stochastic Measurement mLSM for the flow over a NACA 4412 airfoil. By using a combination of Particle Image Velocimetry PIV and multiple surface pressure measurements, processed through a POD/mLSM algorithm, estimates of the velocity field from wall pressure alone are extracted. From such estimates knowledge of the state of the flow above the airfoil can be obtained (i.e., attached, fully separated or incipiently separated). Integral to the POD/mLSM algorithm is the measurement estimation of the global POD coefficients. Here we demonstrate for the first time the utility of these time dependent coefficients, which are estimated from surface pressure only, for use in a simple proportional feedback loop (as the time series to drive the actuators) to keep the flow attached. What is particularly exciting is that we are having success in a flow that is turbulent and incipiently separated. This methodology is critical for implementation of realistic feedback flow control since surface measurements and not inflow measurements are required for most practical applications.