In response to the impact of strong vibrations generated by vibratory rollers on driver comfort during construction, a four-degree-of-freedom “vibratory roller-foundation” dynamic model was established. By analyzing the seat vibrations under both grounded and bouncing conditions, it was found that the seat acceleration responses exceeded the defined comfort range. Therefore, a linear quadratic regulator (LQR) control strategy was introduced, and the control parameters were optimized using a particle swarm algorithm. The results indicated that after the implementation of LQR control, the seat acceleration under grounded conditions was significantly reduced, meeting comfort requirements; although improvements were observed under bouncing conditions, the acceleration remained above the “very uncomfortable”range. After particle swarm optimization, the acceleration amplitudes in both grounded and bouncing conditions were further reduced, both conforming to operational comfort standards, thereby providing a theoretical basis for the design and optimization of the vibration reduction system in vibratory rollers.