For the robotic positioning and navigation, visual odometry (VO) system is widely used. However, the errors of the traditional VO accumulate when the robot moves. Besides, this paper proposes a new framework to solve the problem of monocular VO, called MagicVO. Based on the convolutional neural network (CNN) and the bi-directional LSTM (Bi-LSTM), MagicVO outputs a 6-DoF absolute-scale pose at each position of the camera with a sequence of continuous monocular images as input. It does not only utilize the outstanding performance of CNN in extracting the rich features of image frames fully but also learns the geometric relationship from image sequences pre and post through Bi-LSTM to get a more accurate prediction. A pipeline of the MagicVO is shown in this paper. The MagicVO is an end-to-end system, and the results of the experiments on the KITTI and ETH datasets show that MagicVO has a better performance than the traditional VO systems in the accuracy of pose and the generalization ability.
Feature-based visual simultaneous localization and mapping (SLAM) is an effective localization approach for robots in unknown environments. Classic handcrafted features perform well in 2D image matching tasks. However, in the tracking task of SLAM, the region at the edge of the object in the image is often unstable because of the lack of spatial information. In this paper, we refer to the features at the edge of the object as edge-features and propose an effective method to process the edge-features in SLAM named Edge-Feature Razor (EF-Razor) for the above problems. EF-Razor first uses the semantics provided by the object detection YOLOv3 to distinguish edge-features. Through additional constraints on edgefeatures matching in the tracking process, EF-Razor can effectively reduce the impact of unstable features on the SLAM system. Then, EF-Razor adjusts the information matrix to increase the system's trust in the filtered features. This will make the calculation result of the bundle adjustment more stable. In order to evaluate the proposed method, we integrate EF-Razor to ORB-SLAM2 and perform experiments. The comparison results based on public datasets show the proposed method could effectively reduce the absolute trajectory error by 7%. INDEX TERMS edge, features, object detection, simultaneous localization and mapping.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.