Visually impaired people usually find it hard to travel independently in many public places such as airports and shopping malls due to the problems of obstacle avoidance and guidance to the desired location. Therefore, in the highly dynamic indoor environment, how to improve indoor navigation robot localization and navigation accuracy so that they guide the visually impaired well becomes a problem. One way is to use visual SLAM. However, typical visual SLAM either assumes a static environment, which may lead to less accurate results in dynamic environments or assumes that the targets are all dynamic and removes all the feature points above, sacrificing computational speed to a large extent with the available computational power. This paper seeks to explore marginal localization and navigation systems for indoor navigation robotics. The proposed system is designed to improve localization and navigation accuracy in highly dynamic environments by identifying and tracking potentially moving objects and using vector field histograms for local path planning and obstacle avoidance. The system has been tested on a public indoor RGB-D dataset, and the results show that the new system improves accuracy and robustness while reducing computation time in highly dynamic indoor scenes.