Improved SLAM Merged 2D and 3D Sensors for Mobile Robots

Simultaneous Localization and Mapping (SLAM) method has been wildly applied in mobile robots. In order to solve the problem that the acquired location of mobile robot and map of environment are often inaccurate and incomplete. In this paper, the observation from visual sensor is added to the SLAM system which only use 2D sensor (i.e laser rangefinder), and we propose an approach which merges the observation of the visual sensor, odometry and laser rangefinder based on neural network. Because of the wide vision field of visual sensor and the high precision of laser rangefinder, the proposed approach enables the mobile robot acquire more precise estimation of robots pose and accurate map.