Research On Indoor Self-Localization Method Based On Digital Landmark Map

The construction of smart airports and smart cities presents opportunities and challenges for the application of intelligent mobile robots.How to realize autonomous positioning and navigation of mobile robots in a large scene dynamic environment has attracted the attention of many scholars,but low-cost,personalized navigation methods and systems have always been the focus of attention.Therefore,on the issue of indoor personalized navigation in the terminal building,a method for autonomous mapping and localization of indoor digital landmark map(DLM)from the machine perspective is studiedThe main work and contributions of the paper include:Firstly,a mixed feature representation and extraction algorithm for digital landmark map(DLM)in an indoor public environment is proposed.According to the characteristics and peculiarities of the indoor environment of the terminal building,CSURF(Colored Speed Up Robust Features)feature points that combine color invariance,scale invariance and rotation invariance are constructed as mixed feature representation of DLM from the machine perspective,so that the interior digital landmark features of the terminal have a high degree of robustness for environmental changes such as lighting,angle and distance.So that the characteristics of the indoor DLM of the terminal have a high degree of robustness for environmental changes such as lighting,angle and distance.Secondly,a method for autonomous creation of digital landmark maps combining color mutual information entropy MI(Mutual Information)and CSURF features is proposed.Under the RGB-D camera model,the localization of the camera is estimated based on the CSURF feature points to obtain DLM location data.By combining the color information entropy with the CSURF feature,the redundant frames under the similarity of the global color measurement and local CSURF features in the video image sequence are removed,so that the extracted DLM stakes into account the uniqueness of the image feature and the continuity of the visual scene,thus completing the mapping of the indoor DLM.Thirdly,the indoor localization algorithm based on DLM is studied.By constructing bag-of-words model and loop closure algorithm of the improved strategy,the cumulative error of the mobile robot under long-term motion is eliminated,and the pose graph optimization algorithm is used to reduce the impact of environmental noise,simplify the pose calculation work,and improve the real-time performance of the localization algorithm.Finally,experiments are carried out to verify the validity of the localization algorithm based on the indoor DLM.The indoor self-localization experiments and data analysis were carried out in EuRoC MAV data set and in the actual environment respectively.The experimental results show that the self-localization algorithm based on the indoor DLM can meet the requirements of localization accuracy and real-time performance.