The main challenge in the automation of the large rotary crane with tower-torsion is the accurate positioning and vibration suppression of the load-sway. The start-of-the-art optimal trajectory generation approaches need to consider several state and input constraints to increase the accuracy; therefore, it requires a large amount of computation time and is not applicable for the practical environment. This study presents an efficient method for optimal trajectory generation considering load-sway suppression and collision avoidance in a fast computation time, which includes two control strategies: the offline bi-objective trajectory generation between the contradictory objectives of total motion time and the collision avoidance fitting function, and the online modification of the optimal trajectory, which is formulated as one-degree-of-freedom optimization to reduce the total motion time and satisfy the entire constraints. The experimental validation with a lab-scale three-dimensional rotary crane is provided to show the effectiveness of the proposed method for practical applications.