Abstract:Simultaneous localization and mapping (SLAM) based on RGB-D cameras has been widely used for robot localization and navigation in unknown environments. Most current SLAM methods are constrained by static environment assumptions and perform poorly in real-world dynamic scenarios. To improve the robustness and performance of SLAM systems in dynamic environments, this paper proposes a new RGB-D SLAM method for indoor dynamic scenes based on object detection. The method presented in this paper improves on the ORB-SLAM3 framework. First, we designed an object detection module based on YOLO v5 and relied on it to improve the tracking module of ORB-SLAM3 and the localization accuracy of ORB-SLAM3 in dynamic environments. The dense point cloud map building module was also included, which excludes dynamic objects from the environment map to create a static environment point cloud map with high readability and reusability. Full comparison experiments with the original ORB-SLAM3 and two representative semantic SLAM methods on the TUM RGB-D dataset show that: the method in this paper can run at 30+fps, the localization accuracy improved to varying degrees compared to ORB-SLAM3 in all four image sequences, and the absolute trajectory accuracy can be improved by up to 91.10%. The localization accuracy of the method in this paper is comparable to that of DS-SLAM, DynaSLAM and the two recent target detection-based SLAM algorithms, but it runs faster. The RGB-D SLAM method proposed in this paper, which combines the most advanced object detection method and visual SLAM framework, outperforms other methods in terms of localization accuracy and map construction in a dynamic indoor environment and has a certain reference value for navigation, localization, and 3D reconstruction.

Dynamic Visual Localization and Tracking Method Based on Rgb-D Information

RGB‐D SLAM with Moving Object Tracking in Dynamic Environments

Feature Regions Segmentation Based RGB-D Visual Odometry in Dynamic Environment

Research on RGB-D Visual SLAM Algorithm Based on Adaptive Target Detection

Robust 3D Reconstruction with an RGB-D Camera

RGB-T Long-Term Tracking Algorithm Via Local Sampling and Global Proposals

Dynamic RGB-D SLAM Based on Static Probability and Observation Number

Real-time human detection and tracking in complex environments using single RGBD camera

Real-time SLAM Algorithm Based on RGB-D Data

Vision Odometer Based on RGB-D Camera

RGB-D SLAM in Dynamic Environments Using Point Correlations

Single-scale Siamese Network Based RGB-D Object Tracking with Adaptive Bounding Boxes

A Dynamic Scene Vision SLAM Method Incorporating Object Detection and Object Characterization

Detecting and tracking people in real time with RGB-D camera

RGB-D Vision Navigation Via Motion Saliency Measurement and Twin Reprojection Optimization in Complex Dynamic Scenes

Multi-person Detecting and Tracking Based on RGB-D Sensor for a Robot Vision System.

People tracking and following using multiple RGB-D cameras

Real-Time Multi-Modal People Detection and Tracking of Mobile Robots with A RGB-D Sensor.

RGB-D SLAM Algorithm Based on Clustering and Geometric Residuals in Dynamic Environment

Removing Dynamic 3D Objects from Point Clouds of a Moving RGB-D Camera.

A Robust Multi-Object Tracking SLAM System in Dynamic Scenes for Stereo and RGB-D Cameras