This paper addresses the problem of human-robot cooperative object transportation, taking into account human intervention while tracking the desired trajectory. To solve this problem, an object dynamic model during the human-robot transportation is firstly conducted. Then a PID controller is implemented to track the desired contact force and human's position, while other dimensions was moved by a zero-stiffness admittance control framework to facilitate human manipulation. To enhance real-time trajectory correction, a model-predictive controller (MPC) is employed to guide the human to follow the reference trajectory. Meanwhile, an acceleration adjusting factor is designed to switch the motion mode between active interaction and trajectory tracking. Finally, the performance of the proposed approach is verified by a set of experiments under loaded and unloaded transportation conditions.