Robust RGB-D visual odometry based on edges and points

Yao, Erliang; Zhang, Hexin; Xu, Hui; Song, Hwangjun; Zhang, Guoliang

doi:10.1016/j.robot.2018.06.009

Cited by 20 publications

(17 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The RMSE values of the translational drift and rotational drift are shown in Table 2. And from the analyses and discussion in [27], it is clear that the EP-BASED VO achieves the state-of-the-art results. However, our proposed approach integrated with ORB SLAM2 outperforms the other two approaches for most dynamic sequences, which can be found in the Table 2.…”

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 94%

“…And an iterative closest point (ICP) algorithm is implemented by adding the information composed of static-point's weight, intensity weight and geometric weight. Yao et al [27] proposed a real-time visual odometry combining the sparse edge alignment and minimizing reprojection error to obtain robust state estimation. In addition, the edge with the large reprojection error in the dynamic area will be discarded.…”

Section: Related Workmentioning

confidence: 99%

“…The detailed experimental results on the TUM dataset are provided. The work of [27] is used for comparison, which is EP-BASED VO for short, a VO system designed to handle the situations in dynamic environments. Fig.…”

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 99%

“…Finally, a complete comparison between EP-based VO and our proposed approach is listed in TABLE 3. From [27], it can be known that the original ORB SLAM2 performs worse than the EP-based VO. However, the ORB SLAM2 with our approach is more stable and robust.…”

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 99%

See 3 more Smart Citations

A Compatible Framework for RGB-D SLAM in Dynamic Scenes

et al. 2019

View full text Add to dashboard Cite

Localization and mapping in a dynamic scene is a crucial problem for the indoor visual simultaneous localization and mapping (SLAM) system. Most existed visual odometry (VO) or SLAM systems are based on the assumption that the environment is static. The performance of a SLAM system may degenerate when it is operated in a severely dynamic environment. The assumption limits the applications of RGB-D SLAM in the dynamic environment. In this paper, we propose a workflow to segment the objects accurately, which will be marked as the potentially dynamic-object area based on the semantic information. A novel approach for motion detection and removal from the moving camera is introduced. We integrate the semantics-based motion detection and the segmentation approach with an RGB-D SLAM system. To evaluate the effectiveness of the proposed approach, we conduct the experiments on the challenging dynamic sequences of TUM-RGBD datasets. The experimental results suggest that our approach improves the accuracy of localization and outperforms the state-of-the-art dynamic-removal-based SLAM system in both severely dynamic and slightly dynamic scenes.INDEX TERMS SLAM, dynamic environment, image segmentation.

show abstract

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 94%

Section: Related Workmentioning

confidence: 99%

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 99%

Section: B Comparison With State-of-the-art Vo Systemmentioning

confidence: 99%

See 2 more Smart Citations

A Compatible Framework for RGB-D SLAM in Dynamic Scenes

et al. 2019

View full text Add to dashboard Cite

show abstract

“…One of the most used methods is Simultaneous Localization and Mapping (SLAM) [2] where the environment is not known, so a vision based system is placed on a robot to observe the area and create a map of it, eventually to perform trajectory planning on top of it. There were many enhancements of the SLAM methods proposed, some of them are compared by Santos [3] and the most common sensors used there are lidars [4][5][6][7] and depth cameras [4,[8][9][10]. However, when the environment is known (indoor areas frequently), the advanced SLAM systems may become redundant and if more robots are deployed, it becomes costly when every single robot has to carry all sensors.…”

Section: Introductionmentioning

confidence: 99%

Smart Building Surveillance System as Shared Sensory System for Localization of AGVs

et al. 2020

View full text Add to dashboard Cite

The objective of this study is to extend the possibilities of robot localization in a known environment by using the pre-deployed infrastructure of a smart building. The proposed method demonstrates a concept of a Shared Sensory System for the automated guided vehicles (AGVs), when already existing camera hardware of a building can be utilized for position detection of marked devices. This approach extends surveillance cameras capabilities creating a general sensory system for localization of active (automated) or passive devices in a smart building. The application is presented using both simulations and experiments for a common corridor of a building. The advantages and disadvantages are stated. We analyze the impact of the captured frame’s resolution on the processing speed while also using multiple cameras to improve the accuracy of localization. The proposed methodology in which we use the surveillance cameras in a stand-alone way or in a support role for the AGVs to be localized in the environment has a huge potential utilization in the future smart buildings and cities. The available infrastructure is used to provide additional features for the building control unit, such as awareness of the position of the robots without the need to obtain this data directly from the robots, which would lower the cost of the robots themselves. On the other hand, the information about the location of a robot may be transferred bidirectionally between robots and the building control system to improve the overall safety and reliability of the system.

show abstract

Vision sensor‐based SLAM problem for small UAVs in dynamic indoor environments

Zhou

Kong

Liu

et al. 2022

Computer Animation & Virtual

View full text Add to dashboard Cite

In recent years, with the rapid development of artificial intelligence, machine vision and other related technologies, there has been a demand for higher levels of intelligence in UAVs. There are many excellent SLAM systems available, but most of them assume that their working environment is static. When there are dynamic objects in the environment the localization and mapping accuracy of the SLAM system is reduced, and it can even cause its tracking to fail. To solve this problem. In this paper, we propose a target detection-based SLAM algorithm for real-time operation in crowded dynamic environments.The algorithm removes feature points of dynamic objects from key frames and then constructs a point cloud map of based on the state-of-the-art SLAM system ORB-SLAM3. Finally, the proposed method is validated and evaluated on multiple dynamic datasets and real environments. The results show that the algorithm in this paper outperforms other visual SLAM algorithms in terms of localization and map building accuracy in more highly dynamic datasets, while guaranteeing performance.

show abstract

Robust RGB-D visual odometry based on edges and points

Cited by 20 publications

References 24 publications

A Compatible Framework for RGB-D SLAM in Dynamic Scenes

A Compatible Framework for RGB-D SLAM in Dynamic Scenes

Smart Building Surveillance System as Shared Sensory System for Localization of AGVs

Vision sensor‐based SLAM problem for small UAVs in dynamic indoor environments

Contact Info

Product

Resources

About