Aiming at the problem of incomplete scanning map information and poor anti-interference ability when the robot uses a single sensor for fast simultaneous locating and mapping (FastSLAM), a novel method based on multiple sensor information fusion is proposed. At the first, the RGB-D vision sensor, laser range finder and odometer are preprocessed by noise reduction, and the depth image of RGB-D vision sensor is converted into the simulated laser data. The feature information is acquired by the simulated laser data and laser sensor data. The simulate laser and the real-time laser can be fused by introduced information fusion rule in this paper, this step expands the robot's vision and makes up for the visual blind spots of the robot single sensor, and obtaining a high-accurate map. Then, the unscented Kalman filter (UKF) algorithm is introduced in FastSLAM method, the new particles proposal distribution can be obtained. This step can not only avoids increasing the complexity of the algorithm when calculating the Jacobi matrix, but also improve the system's state estimation accuracy and the speed of algorithm convergence. Finally, the real scene experiments on the method of this paper are carried out on a Pioneer3-DX robot equipped with robot operating system (ROS), laser sensor and RGB-D camera. It is proved that this paper proposes a fast simultaneous locating and mapping method based on information fusion of RGB-D and laser sensor can not only improve the accuracy of the SLAM, but also enable the robot to construct a high-precision grid map in the environment.