Research On SLAM System Of Indoor Mobile Robot

With the continuous development of artificial intelligence technology,people's demand for mobile robots is getting higher and higher.Therefore,building an intelligent and autonomous mobile robot becomes very research value and market prospect.Simultaneous Localization and Mapping(SLAM)is considered to be a key technology in the field of robotics research.It solves the shortcomings of laser SLAM,which is complicated,expensive and professional.Therefore,it has strong research significance and value.To this end,the SLAM system of indoor mobile robot based on vision will be built in this article.Firstly,the camera parameters are calibrated by Zhang's calibration method,and ORB algorithm with better selectivity is used as feature extraction algorithm.RANSAC algorithm is used to eliminate mismatches,and the position and pose of the camera are calculated by polar geometric positioning algorithm based on the correct matching point pairs.Secondly,using the positioning results and different data types,a sparse and dense point cloud map is constructed to meet the needs of different scenarios.Then,aiming at the influence of noise and positioning error,a non-linear least squares optimization model is constructed,and the whole optimization model is represented by a graph using graph theory.The optimization solution is carried out by using open source G2 O framework.Finally,the whole code is synthesized to complete the construction of visual SLAM system.In order to verify the positioning and mapping effect of the SLAM system in this article,TUM data sets and real-time environment data are used to experiment.The results show that the SLAM system in this article satisfies both positioning accuracy and real-time performance,and the dense map generated offline conforms to the expected experimental results.