Deep Reinforcement Learning (DRL) has emerged as a promising control strategy, and its application in the Motion Cueing Algorithm (MCA) for driving simulators has gained validation. However, challenges persist, including low sampling efficiency, significant online learning time, and instability during application, which hinder its widespread adoption in generic driving simulator MCAs. This study introduces an MCA strategy based on Model Predictive Control (MPC) augmented Reinforcement Learning (RL). By directly integrating MPC with RL to guide the training process, this method reduces oscillations in the convergence curve of reinforcement learning, enabling faster convergence and possessing learning capabilities that MPC-MCA lacks once convergence is achieved. Numerical simulations conducted on Renault’s 8DOF simulator confirm that the proposed method outperforms the current state-of-the-art RL-based MCA, significantly reducing the generation of contradictory false motion cue signals. The proposed method paves the way for the broader application of DRL in general MCAs.