Abstract
A visual localization approach for dynamic objects based on hybrid semantic-geometry information is presented. Due to the interference of moving objects in the real environment, the traditional simultaneous localization and mapping (SLAM) system can be corrupted. To address this problem, we propose a method for static/dynamic image segmentation that leverages semantic and geometric modules, including optical flow residual clustering, epipolar constraint checks, semantic segmentation, and outlier elimination. We integrated the proposed approach into the state-of-the-art ORB-SLAM2 and evaluated its performance on both public datasets and a quadcopter platform. Experimental results demonstrated that the root-mean-square error of the absolute trajectory error improved, on average, by 93.63% in highly dynamic benchmarks when compared with ORB-SLAM2. Thus, the proposed method can improve the performance of state-of-the-art SLAM systems in challenging scenarios.
Highlights
Experimental results demonstrated that the root-mean-square error of the absolute trajectory error improved, on average, by 93.63% in highly dynamic benchmarks when compared with ORB-SLAM2
We propose a novel approach to the elimination of dynamic points, which can greatly improve the feasibility and effectiveness of the state-of-the-art ORBSLAM2 system
We developed an outliers rejection strategy to allow our system to deal with dynamic environments
Summary
LSD-SLAM [6] maintains highly accurate pose estimation based on direct image alignment, as well as reconstruction of the 3D environment as a pose-graph of keyframes with associated semi-dense depth maps. Detect-SLAM [19] incorporates a deep neural network (DNN) object detector into the SLAM system to benefit from these two mutually beneficial functions This system categorizes keypoints into four states according to their moving probability, and removes all points with a high moving probability to maintain a robust pose estimation. As the front-end of the ORBSLAM2, this method significantly improves the precision of the state-of-the-art SLAM system in various challenging scenarios This learning-based method can identify dynamic objects without the need of multi-frame processing. Compared with state-of-the-art SLAM (e.g., ORBSLAM2) and other prominent dynamic SLAM systems, our approach demonstrated superior intelligence in challenging scenarios
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call