Learning to See the Wood for the Trees: Deep Laser Localization in Urban and Natural Environments on a CPU

Tinchev, Georgi; Peñate-Sánchez, Adrián; Fallon, Maurice

doi:10.1109/lra.2019.2895264

Cited by 46 publications

(51 citation statements)

References 25 publications

(59 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For data association between map landmarks and lidar data, Kukko et al (2017) used simple criteria such as diameter similarity and distance. Tinchev, Penate‐Sanchez, and Fallon (2019) further explored this data association problem, but made use of deep learning to detect loop closures while mapping forests with lidar; this could prove more robust than simply comparing the diameter of trees for data association. Garforth and Webb (2019) are the only ones who tested monocular vision‐based SLAM such as ORBSLAM in forests, mostly coming to the conclusion that more research is needed.…”

Section: Introductionmentioning

confidence: 99%

Automatic three‐dimensional mapping for tree diameter measurements in inventory operations

Tremblay

Béland

Gagnon

et al. 2020

Journal of Field Robotics

View full text Add to dashboard Cite

Forestry is a major industry in many parts of the world, yet this potential domain of application area has been overlooked by the robotics community. For instance, forest inventory, a cornerstone of efficient and sustainable forestry, is still traditionally performed manually by qualified professionals. The lack of automation in this particular task, consisting chiefly of measuring tree attributes, limits its speed, and, therefore, the area that can be economically covered. To this effect, we propose to use recent advancements in three-dimensional mapping approaches in forests to automatically measure tree diameters from mobile robot observations. While previous studies showed the potential for such technology, they lacked a rigorous analysis of diameter estimation methods in challenging and large-scale forest environments. Here, we validated multiple diameter estimation methods, including two novel ones, in a new publicly-available dataset which includes four different forest sites, 11 trajectories, totaling 1458 tree observations, and 14,000 m 2. From our extensive validation, we concluded that our mapping method is usable in the context of automated forest inventory, with our best diameter estimation method yielding a root mean square error of 3.45 cm for our whole dataset and 2.04 cm in ideal conditions consisting of mature forest with well-spaced trees. Furthermore, we release this dataset to the public (https://norlab.ulaval.ca/research/montmorencydataset), to spur further research in robotic forest inventories. Finally, stemming from this large-scale experiment, we provide recommendations for future deployments of mobile robots in a forestry context.

show abstract

Section: Introductionmentioning

confidence: 99%

Automatic three‐dimensional mapping for tree diameter measurements in inventory operations

Tremblay

Béland

Gagnon

et al. 2020

Journal of Field Robotics

View full text Add to dashboard Cite

show abstract

“…The localization of the vehicle is obtained through convolutional matching. In Tinchev, Penate‐Sanchez, and Fallon (), laser scans and a deep neural network are used to learn descriptors for localization in urban and natural environments.…”

Section: Deep Learning For Driving Scene Perception and Localizationmentioning

confidence: 99%

A survey of deep learning techniques for autonomous driving

Grigorescu

Trasnea

Cocias

et al. 2019

Journal of Field Robotics

1,127

495

View full text Add to dashboard Cite

The last decade witnessed increasingly rapid progress in self‐driving vehicle technology, mainly backed up by advances in the area of deep learning and artificial intelligence (AI). The objective of this paper is to survey the current state‐of‐the‐art on deep learning technologies used in autonomous driving. We start by presenting AI‐based self‐driving architectures, convolutional and recurrent neural networks, as well as the deep reinforcement learning paradigm. These methodologies form a base for the surveyed driving scene perception, path planning, behavior arbitration, and motion control algorithms. We investigate both the modular perception‐planning‐action pipeline, where each module is built using deep learning methods, as well as End2End systems, which directly map sensory information to steering commands. Additionally, we tackle current challenges encountered in designing AI architectures for autonomous driving, such as their safety, training data sources, and computational hardware. The comparison presented in this survey helps gain insight into the strengths and limitations of deep learning and AI approaches for autonomous driving and assist with design choices.

show abstract

“…The feature extraction network Darknet53 of the YOLOv3 real-time target detection model is composed of 53 convolutional layers and 24 residual layers [27], [28]. The last 20 layers are the feature interaction layer of the YOLO network, which is divided into 3 scales.…”

Section: A Object Detection Using Improved Yolov3mentioning

confidence: 99%

3D Semantic Map Construction Using Improved ORB-SLAM2 for Mobile Robot in Edge Computing Environment

Cui

Wang

2020

IEEE Access

View full text Add to dashboard Cite

Although the existing localization and mapping (SLAM) technology of indoor mobile robot has made great development, its intelligence and environmental perception ability still cannot meet the needs of service and inspection. Therefore, based on edge computing environment, a 3D semantic map construction of mobile robot based on improved ORB-SALM2 is proposed. Firstly, the improved yolov3 algorithm is used to detect indoor objects, and then the real-time semantic segmentation network model based on deep learning is used to segment indoor objects to achieve the classification of pixel points of objects on two-dimensional images, and BAFF feature fusion algorithm is introduced to improve the accuracy of semantic segmentation model. Then, through the SLAM system, we estimate the pose of the image in the result of semantic segmentation, and use the depth information to project it into the three-dimensional environment to build the three-dimensional semantic map. Finally, the experiment platform of mobile robot is built to verify the stability of ORB-D and thermal imaging sensor registration technology, the accuracy and real-time of building three-dimensional environment thermal field map, and the accuracy of robot positioning using thermal infrared and depth image.

show abstract

Learning to See the Wood for the Trees: Deep Laser Localization in Urban and Natural Environments on a CPU

Cited by 46 publications

References 25 publications

Automatic three‐dimensional mapping for tree diameter measurements in inventory operations

Automatic three‐dimensional mapping for tree diameter measurements in inventory operations

A survey of deep learning techniques for autonomous driving

3D Semantic Map Construction Using Improved ORB-SLAM2 for Mobile Robot in Edge Computing Environment

Contact Info

Product

Resources

About