The advancement of robot navigation holds significant importance across diverse domains. However, traditional methodologies predominantly reliant on single-sensor systems result in insufficient perception of the environment, thus limiting the efficacy of autonomous navigation. This paper proposes a novel deep reinforcement learning method that leverages sensor fusion and shared buffer replay to enhance robot navigation capabilities. By fusing depth images and LiDAR data, the proposed approach enhances the robot’s understanding of its environment, enabling precise decision-making within complex terrains. Furthermore, the incorporation of shared buffer replay allows for simultaneous training of multiple robots in simulated environments, expediting the acquisition of optimal navigation strategies. Experimental evaluations underscore the efficacy of the proposed method, demonstrating its achievement of state-of-the-art performance metrics including success rate, collision rate, timeout rate, navigation time, and trajectory length.