OverlapNet: Loop Closing for LiDAR-based SLAM

Chen, Xieyuanli; Läbe, Thomas; Milioto, Andres; Röhling, Timo; Vysotska, Olga; Haag, Alexandre; Behley, Jens; Stachniss, Cyrill

doi:10.15607/rss.2020.xvi.009

Cited by 188 publications

(127 citation statements)

References 35 publications

Supporting

Mentioning

127

Contrasting

Order By: Relevance

“…Closed-loop detection is an important part of the SLAM system and plays a very important role in eliminating accumulated errors. There are many references to the closed-loop detection, for example [ 27 , 28 ]. Considering the detection speed, this paper adopted the fast closed-loop detection method mentioned in the literature [ 19 ].…”

Section: Methodsmentioning

confidence: 99%

Continuous-Time Laser Frames Associating and Mapping via Multilayer Optimization

Xiao

Zhang

et al. 2020

Sensors

View full text Add to dashboard Cite

To achieve the ability of associating continuous-time laser frames is of vital importance but challenging for hand-held or backpack simultaneous localization and mapping (SLAM). In this study, the complex associating and mapping problem is investigated and modeled as a multilayer optimization problem to realize low drift localization and point cloud map reconstruction without the assistance of the GNSS/INS navigation systems. 3D point clouds are aligned among consecutive frames, submaps, and closed-loop frames using the normal distributions transform (NDT) algorithm and the iterative closest point (ICP) algorithm. The ground points are extracted automatically, while the non-ground points are automatically segmented to different point clusters with some noise point clusters omitted before 3D point clouds are aligned. Through the three levels of interframe association, submap matching and closed-loop optimization, the continuous-time laser frames can be accurately associated to guarantee the consistency of 3D point cloud map. Finally, the proposed method was evaluated in different scenarios, the experimental results showed that the proposed method could not only achieve accurate mapping even in the complex scenes, but also successfully handle sparse laser frames well, which is critical for the scanners such as the new Velodyne VLP-16 scanner’s performance.

show abstract

Section: Methodsmentioning

confidence: 99%

Continuous-Time Laser Frames Associating and Mapping via Multilayer Optimization

Xiao

Zhang

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

Section: Hierarchical Localisationmentioning

confidence: 99%

“…The work in Reference [34] is related to ours in the sense that it can also be seen as a hierarchical approach to localisation, in the particular case of LiDAR. We see two major distinctions between Reference [34] and our work. Firstly, to achieve rotational invariance we account for the cylindrical nature of the scan formation process in our architecture, while Reference [34] maintains the rotation information and exploits it to extract the yaw displacement from the comparison of the embeddings of two scans.…”

Section: Hierarchical Localisationmentioning

confidence: 99%

“…Firstly, to achieve rotational invariance we account for the cylindrical nature of the scan formation process in our architecture, while Reference [34] maintains the rotation information and exploits it to extract the yaw displacement from the comparison of the embeddings of two scans. Secondly, in Reference [34], the measure of overlap for a pair of scans is a nonlinear, complex function that takes both embeddings as input rather than a simple distance measure. As such, the query procedure prevents the direct use of efficient searching procedures, such as the k-D-trees we use in our RPR module.…”

Section: Hierarchical Localisationmentioning

confidence: 99%

See 1 more Smart Citation

kRadar++: Coarse-to-Fine FMCW Scanning Radar Localisation

Martini

Gadd

Newman

2020

Sensors

View full text Add to dashboard Cite

This paper presents a novel two-stage system which integrates topological localisation candidates from a radar-only place recognition system with precise pose estimation using spectral landmark-based techniques. We prove that the—recently available—seminal radar place recognition (RPR) and scan matching sub-systems are complementary in a style reminiscent of the mapping and localisation systems underpinning visual teach-and-repeat (VTR) systems which have been exhibited robustly in the last decade. Offline experiments are conducted on the most extensive radar-focused urban autonomy dataset available to the community with performance comparing favourably with and even rivalling alternative state-of-the-art radar localisation systems. Specifically, we show the long-term durability of the approach and of the sensing technology itself to autonomous navigation. We suggest a range of sensible methods of tuning the system, all of which are suitable for online operation. For both tuning regimes, we achieve, over the course of a month of localisation trials against a single static map, high recalls at high precision, and much reduced variance in erroneous metric pose estimation. As such, this work is a necessary first step towards a radar teach-and-repeat (RTR) system and the enablement of autonomy across extreme changes in appearance or inclement conditions.

show abstract

“…Most semantic SLAM methods fuse semantic labels obtained from semantic segmentation and maps generated by the SLAM algorithm to generate 3D maps with semantic information. According to the type of sensor they used, semantic SLAM algorithms can be classified as the monocular camerabased [17]- [19], stereo camera-based [20], [21], LiDARbased [22]- [24], multiple sensors-based [25], [26], RGB-D camera-based approaches, and so on. This paper mainly considers semantic SLAM algorithms based on the RGB-D camera.…”

Section: Semantic Slammentioning

confidence: 99%

Semantic RGB-D SLAM for Rescue Robot Navigation

et al. 2020

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a semantic simultaneous localization and mapping (SLAM) framework for rescue robots, and report its use in navigation tasks. Our framework can generate not only geometric maps in the form of dense point-clouds but also corresponding point-wise semantic labels generated by a semantic segmentation convolutional neural network (CNN). The semantic segmentation CNN is trained using our RGB-D dataset of the RoboCup Rescue-Robot-League (RRL) competition environment. With the help of semantic information, the rescue robot can identify different types of terrains in a complex environment, so as to avoid specific obstacles or to choose routes with better traversability. To reduce the segmentation noise, our approach utilizes depth images to perform filtering on the segmentation results of each frame. The overall semantic map is then further improved in the point-cloud voxels. By accumulating results of multiple frames in the voxels, semantic maps with consistent semantic labels are obtained. To show the advantage of having a semantic map of the environment, we report a case study of how the semantic map can be utilized in a navigation task to reduce the arrival time while ensuring safety. The experimental result shows that our semantic SLAM framework is capable of generating a dense semantic map for the complex RRL competition environment, with which the arrival time of the navigation time is effectively reduced.

show abstract

OverlapNet: Loop Closing for LiDAR-based SLAM

Cited by 188 publications

References 35 publications

Continuous-Time Laser Frames Associating and Mapping via Multilayer Optimization

Continuous-Time Laser Frames Associating and Mapping via Multilayer Optimization

kRadar++: Coarse-to-Fine FMCW Scanning Radar Localisation

Semantic RGB-D SLAM for Rescue Robot Navigation

Contact Info

Product

Resources

About