Research On SLAM Algorithm Based On Depth Perception Of Indoor Mobile Robots

With the rapid development of computer technology and sensor technology, people on the intelligence level of robot have become increasingly demanding, and use the mobile robot to explore the unknown environment is a hot and difficult research of intelligent robot. Mobile robot simultaneous localization and mapping(SLAM) is one of important basic issues, effectively solve the SLAM problem is considered to be the key to realizing the intelligence of mobile robots. In recent years, with the development of computer vision, visual sensor is gradually applied to the SLAM problem, therefore, the visual SLAM rise gradually becomes an important research direction in the field of SLAM.In this paper, using a new vision sensor Kinect as the depth of visual sensor for indoor mobile robot vision SLAM algorithm was studied, aiming at the disadvantage of traditional laser scanner in map building, proposed an algorithm for 2D SLAM which based on the Kinect depth data simulation of laser scanning data; and at the same time, in order to make up for the application of two-dimensional map simple narrow problem is proposed a 3D SLAM method for mobile robot based on Kinect, in order to combine with the 2D map and greatly expanded its scope of application. The main work is shown as follows*:Firstly, this paper introduces the main structure of Kinect sensor used in this research and the relevant principle, combined with Power BotTM mobile robot experimental platform unified; While the definition of the mobile robot SLAM problem in the coordinate system model and environment map model, and explains the create two-dimensional maps of the raster map representation method for the subsequent experimental results set a unified mobile robot coordinate system model and map representation.Secondly, in the process of map creation, the traditional laser scanner can only detect the obstacles in a specific plane, and we have proposed an algorithm for 2D map creation in unknown environment, which based on the Kinect depth data simulation of laser scanning data. The proposed method through the depth image and color image acquisition Kinect indoor environment sensor, point cloud data obtained by simulation method and the corresponding laser scanning, At the same time, this paper also uses the a Kinect laser scanner simple analog simulation, in order to reduce the simulation time conversion, thereby increasing the map created in real time, finally succeeded in establishing a more detailed, more close to the real environment of 2D map.Finally, the 2D map information possessed by a single relatively narrow range of applications cause problems, we propose a method of creating 3D map based on Kinect, to create a 3D map consistent with the creation of a 2D map. The proposed method of color image and depth image acquisition using Kinect sensor for indoor scenes, the use of SIFT algorithm for color scene graph of feature extraction and matching and GTM algorithm to eliminate the wrong matching points, combined with the depth information of the scene to guide the ICP algorithm for interframe registration of 3D point cloud, and out of two adjacent rotation matrix and translation matrix, according to the two matrix into 3D map image sequence obtained during operation, so as to create a more detailed three-dimensional map of the indoor environment.