Research On Simultaneous Localization And Mapping In Indoor Robot Based On Vision

With the rapid development of computer technology and sensor technology,people on the intelligence level of robot h ave 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 considered to be the key to realizing the inte lligence 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 gradually becomes an important research direction in the field of SLAM.Current RGB-D SLAM algorithm includes problems as below: first,it has such a low efficiency that it cannot meet the real-time requirement;second,it has a low accuracy and a large error,the resulting robot pose and trajectory will generally drift with respect to the real ones and the drift amount grows over time.Aiming at solving the problems mentioned above,this paper proposed the following improvement methods:(1)Use double threshold to detect FAST feature points,and combine four fork tree principle to improve OR B feature in the phase of feature detection and descriptor extraction,use improved ORB algorithm to detect feature and extract descriptors.(2)the matching results are optimized by the removal of mismatching method based on the fusion of GMS and characteristic direction in the phase of feature matching.(3)In the front-end of SLAM,computational cost is applied.In this paper,Grid acceleration is applied to feature extraction and matching steps,so as to improve accuracy,while the algorithm speed is not reduced.This paper uses a benchmark and a corresponding dataset proposed by TUM and some RGB-D images sequences of different environments recorded by ourselves and the real environment based on Turtle Bot2 robot to evaluate the original algorithm and the improved algorithm,the result show that the improvement methods of the algorithm proposed in this paper can not only meet the real-time requirement,but also greatly reduce the errors and improve the accuracy.This proves the correctness of the improvement method proposed in this paper.