Test Your SLAM! The SubT-Tunnel dataset and metric for mapping

Rogers, John G.; Gregory, Jason M.; Fink, Jonathan; Stump, Ethan

doi:10.1109/icra40945.2020.9197156

Cited by 35 publications

(19 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, using the quadrupedal robot ANYmal, the suitability of the proposed approach for real-world autonomous missions is demonstrated by integrating its pose estimates into a mapping pipeline and comparing against a state-ofthe-art model-based approach [3]. Next, reliability of the proposed approach is demonstrated by applying it to datasets from the DARPA SubT Challenge [11], collected using a tracked robot, and comparing the built map against the ground-truth map. Finally, to aid numerical comparison with existing work, an evaluation is conducted on the KITTI odometry benchmark [33].…”

Section: Resultsmentioning

confidence: 99%

“…Next, the proposed approach is tested on the DARPA SubT Challenge Urban Circuit datasets [11]. These datasets were collected using an iRobot PackBot Explorer tracked robot carrying an Ouster OS1-64 LiDAR at Satsop Business Park in Washington, USA.…”

Section: B Darpa Subt Challenge Urban Circuitmentioning

confidence: 99%

“…The performance of the proposed approach is verified and compared against existing methods on driving datasets. Furthermore, the suitability towards complex real-world robotic applications is demonstrated for the first time by conducting autonomous mapping missions with the quadrupedal robot ANYmal [10], shown in operation in Figure 1, as well as evaluating the mapping performance on DARPA Subterranean (SubT) Challenge datasets [11]. Finally, the code of the proposed method is publicly available for the benefit of the robotics community 1 .…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Self-supervised Learning of LiDAR Odometry for Robotic Applications

Nubert

Khattak

Hutter

2021

2021 IEEE International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

Reliable robot pose estimation is a key building block of many robot autonomy pipelines, with LiDAR localization being an active research domain. In this work, a versatile self-supervised LiDAR odometry estimation method is presented, in order to enable the efficient utilization of all available LiDAR data while maintaining real-time performance. The proposed approach selectively applies geometric losses during training, being cognizant of the amount of information that can be extracted from scan points. In addition, no labeled or ground-truth data is required, hence making the presented approach suitable for pose estimation in applications where accurate ground-truth is difficult to obtain. Furthermore, the presented network architecture is applicable to a wide range of environments and sensor modalities without requiring any network or loss function adjustments. The proposed approach is thoroughly tested for both indoor and outdoor real-world applications through a variety of experiments using legged, tracked and wheeled robots, demonstrating the suitability of learningbased LiDAR odometry for complex robotic applications.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: B Darpa Subt Challenge Urban Circuitmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Self-supervised Learning of LiDAR Odometry for Robotic Applications

Nubert

Khattak

Hutter

2021

2021 IEEE International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

show abstract

“…The vision of having a sustainable swarm of autonomous aerial vehicles [ 9 ] has attracted researcheres from both academy and industry [ 10 , 11 , 12 , 13 ]. Yet, even the concept of a single autonomous drone still encapsulates a wide range of challenges [ 14 , 15 ]. Recent improvement in hardware and software for edge deep learning platforms [ 16 , 17 ] allows micro-drones to use visual sensors for obstacle avoidance and navigation [ 18 ].…”

Section: Introductionmentioning

confidence: 99%

Vision-Less Sensing for Autonomous Micro-Drones

Pikalov

Azaria

Sonnenberg

et al. 2021

Sensors

View full text Add to dashboard Cite

This work presents a concept of intelligent vision-less micro-drones, which are motivated by flying animals such as insects, birds, and bats. The presented micro-drone (named BAT: Blind Autonomous Tiny-drone) can perform bio-inspired complex tasks without the use of cameras. The BAT uses LIDARs and self-emitted optical-flow in order to perform obstacle avoiding and maze-solving. The controlling algorithms were implemented on an onboard micro-controller, allowing the BAT to be fully autonomous. We further present a method for using the information collected by the drone to generate a detailed mapping of the environment. A complete model of the BAT was implemented and tested using several scenarios both in simulation and field experiments, in which it was able to explore and map complex building autonomously even in total darkness.

show abstract

“…This makes accessing ground truth data easy, but it introduces new challenges to create realistic virtual environments and camera trajectories. Some datasets [9,10,11] choose to use pseudo-ground truth data which are generated by estimation algorithms, while some others resort to external markers [12,13].…”

Section: Introductionmentioning

confidence: 99%

A Metric For Evaluating 3d Reconstruction And Mapping Performance With No Ground Truthing

Zhang

Chen

2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

It is not easy when evaluating 3D mapping performance because existing metrics require ground truth data that can only be collected with special instruments. In this paper, we propose a metric, dense map posterior (DMP), for this evaluation. It can work without any ground truth data. Instead, it calculates a comparable value, reflecting a map posterior probability, from dense point cloud observations. In our experiments, the proposed DMP is benchmarked against ground truth-based metrics. Results show that DMP can provide a similar evaluation capability. The proposed metric makes evaluating different methods more flexible and opens many new possibilities, such as self-supervised methods and more available datasets.

show abstract

Test Your SLAM! The SubT-Tunnel dataset and metric for mapping

Cited by 35 publications

References 9 publications

Self-supervised Learning of LiDAR Odometry for Robotic Applications

Self-supervised Learning of LiDAR Odometry for Robotic Applications

Vision-Less Sensing for Autonomous Micro-Drones

A Metric For Evaluating 3d Reconstruction And Mapping Performance With No Ground Truthing

Contact Info

Product

Resources

About