Moving Object Segmentation and Detection for Robust RGBD-SLAM in Dynamic Environments

Xie, Wanfang; Liu, Peter; Zheng, Minhua

doi:10.1109/tim.2020.3026803

Cited by 63 publications

(30 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Use small radius side windows for multilayer fusion; (4) end if (5) if l ≥ 1/2L then (6) Use Yolov5 to distinguish static and semistatic objects; (7) Increase the radius of the side windows of the regions where the semistatic objects are detected; (8) end if (9) end for (10) visual odometry and visual SLAM systems. In this paper, the sequences "freiburg1_xyz," "frei-burg2_xyz," "freiburg2_rpy," "freiburg1_desk," and "freiburg1_desk2" are selected, which were all acquired in the office interior scene with rich texture.…”

Section: Experimental Results and Analysismentioning

confidence: 99%

A Variable Radius Side Window Direct SLAM Method Based on Semantic Information

Chen

Mutabazi

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Simultaneous Localization and Mapping (SLAM) is a challenging and key issue in the mobile robotic fields. In terms of the visual SLAM problem, the direct methods are more suitable for more expansive scenes with many repetitive features or less texture in contrast with the feature-based methods. However, the robustness of the direct methods is weaker than that of the feature-based methods. To deal with this problem, an improved direct sparse odometry with loop closure (LDSO) is proposed, where the performance of the SLAM system under the influence of different imaging disturbances of the camera is focused on. In the proposed method, a method based on the side window strategy is proposed for preprocessing the input images with a multilayer stacked pixel blender. Then, a variable radius side window strategy based on semantic information is proposed to reduce the weight of selected points on semistatic objects, which can reduce the computation and improve the accuracy of the SLAM system based on the direct method. Various experiments are conducted on the KITTI dataset and TUM RGB-D dataset to test the performance of the proposed method under different camera imaging disturbances. The quantitative and qualitative evaluations show that the proposed method has better robustness than the state-of-the-art direct methods in the literature. Finally, a real-world experiment is conducted, and the results prove the effectiveness of the proposed method.

show abstract

Section: Experimental Results and Analysismentioning

confidence: 99%

A Variable Radius Side Window Direct SLAM Method Based on Semantic Information

Chen

Mutabazi

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

show abstract

“…While OpenPose can be used to estimate the positions of body joints of humans within an RGB image, other dynamic scene reconstruction methods consider instance segmentation or optical flow techniques to separate dynamic parts of the scene from the static background. For example, Xie et al (2021) perform feature-based RGB-D Simultaneous Localisation and Mapping (SLAM) in dynamic environments using Mask R-CNN (He et al 2020) image segmentation and optical flow-based motion detection. While their approach demonstrates state-of-the-art localisation accuracy, they report an average processing time per frame of 0.42 s and "up to 1.10 s when mask inpainting" is required.…”

Section: Perception and Motion Planning In Dynamic Environmentsmentioning

confidence: 99%

“…As discussed in Sec. 2.2, numerous approaches have been explored in previous efforts to perform dense environment mapping in dynamic environments (Scona et al 2018;Zhang and Nakamura 2020;Cao et al 2021;Palazzolo et al 2019), with many employing image segmentation techniques (Xie et al 2021;Runz et al 2019;Zhang et al 2019).…”

Section: Image Processingmentioning

confidence: 99%

Motion Planning in Dynamic Environments Using Context-Aware Human Trajectory Prediction

Finean¹,

Petrović²,

Merkt³

et al. 2022

Preprint

View full text Add to dashboard Cite

Over the years, the separate fields of motion planning, mapping, and human trajectory prediction have advanced considerably. However, the literature is still sparse in providing practical frameworks that enable mobile manipulators to perform whole-body movements and account for the predicted motion of moving obstacles. Previous optimisationbased motion planning approaches that use distance fields have suffered from the high computational cost required to update the environment representation. We demonstrate that GPU-accelerated predicted composite distance fields significantly reduce the computation time compared to calculating distance fields from scratch. We integrate this technique with a complete motion planning and perception framework that accounts for the predicted motion of humans in dynamic environments, enabling reactive and pre-emptive motion planning that incorporates predicted motions. To achieve this, we propose and implement a novel human trajectory prediction method that combines intention recognition with trajectory optimisation-based motion planning. We validate our resultant framework on a real-world Toyota Human Support Robot (HSR) using live RGB-D sensor data from the onboard camera. In addition to providing analysis on a publicly available dataset, we release the Oxford Indoor Human Motion (Oxford-IHM) dataset and demonstrate stateof-the-art performance in human trajectory prediction. The Oxford-IHM dataset is a human trajectory prediction dataset in which people walk between regions of interest in an indoor environment. Both static and robot-mounted RGB-D cameras observe the people while tracked with a motion-capture system.

show abstract

“…3) Other methods: In [29], the VOS and optical flow are iteratively optimized, where the VOS is realized by a optical flow-based graphical model. In [30], the moving objects are divided into active and passive moving ones according to the optical flow. Then, the passive moving objects can be segmented by a simple calculation.…”

Section: B Optical Flow-based Vosmentioning

confidence: 99%

FAMINet: Learning Real-time Semi-supervised Video Object Segmentation with Steepest Optimized Optical Flow

Liu¹,

Liu²,

Chen³

et al. 2021

Preprint

View full text Add to dashboard Cite

Semi-supervised video object segmentation (VOS) aims to segment a few moving objects in a video sequence, where these objects are specified by annotation of first frame. The optical flow has been considered in many existing semisupervised VOS methods to improve the segmentation accuracy. However, the optical flow-based semi-supervised VOS methods cannot run in real time due to high complexity of optical flow estimation. A FAMINet, which consists of a feature extraction network (F), an appearance network (A), a motion network (M), and an integration network (I), is proposed in this study to address the abovementioned problem. The appearance network outputs an initial segmentation result based on static appearances of objects. The motion network estimates the optical flow via very few parameters, which are optimized rapidly by an online memorizing algorithm named relaxed steepest descent. The integration network refines the initial segmentation result using the optical flow. Extensive experiments demonstrate that the FAMINet outperforms other state-of-the-art semi-supervised VOS methods on the DAVIS and YouTube-VOS benchmarks, and it achieves a good trade-off between accuracy and efficiency. Our code is available at https://github.com/liuziyang123/FAMINet.

show abstract

Moving Object Segmentation and Detection for Robust RGBD-SLAM in Dynamic Environments

Cited by 63 publications

References 26 publications

A Variable Radius Side Window Direct SLAM Method Based on Semantic Information

A Variable Radius Side Window Direct SLAM Method Based on Semantic Information

Motion Planning in Dynamic Environments Using Context-Aware Human Trajectory Prediction

FAMINet: Learning Real-time Semi-supervised Video Object Segmentation with Steepest Optimized Optical Flow

Contact Info

Product

Resources

About