This paper presents a novel approach for controlling parallel kinematic manipulators (PKMs) using a feedforward augmented model reference adaptive control (MRAC) scheme. The original direct MRAC approach lacks the knowledge of the dynamic model and does not ensure boundedness of the feedback gains. To overcome these limitations, our proposed approach incorporates a feedforward dynamic term to enhance the tracking performance, and a projection operator to guarantee the boundedness of the feedback gains. The proposed controller is validated through real-time experiments using a 6degree-of-freedom (DOF) PKM, and is compared with the original direct MRAC and some stateof-the-art controllers in various scenarios, including nominal and robustness cases. The obtained experimental results demonstrate the superiority of the proposed approach in terms of trajectory tracking performances and adaptation efficiency.