353 Hits in 5.4 sec

A fully end-to-end deep learning approach for real-time simultaneous 3D reconstruction and material recognition

Cheng Zhao, Li Sun, Rustam Stolkin
2017 2017 18th International Conference on Advanced Robotics (ICAR)  
Unlike previous methods, we propose a fully end-to-end approach, which does not require hand-crafted features or CRF post-processing.  ...  In contrast, we propose a deep learning method which performs 3D reconstruction while simultaneously recognising different types of materials and labelling them at the pixel level.  ...  Sun was support by RoMaNS. Stolkin was supported by a Royal Society Industry Fellowship.  ... 
doi:10.1109/icar.2017.8023499 dblp:conf/icar/ZhaoSS17 fatcat:qrijis6borhopiykqa535tq35y

Deep Direct Visual Odometry [article]

Chaoqiang Zhao, Yang Tang, Qiyu Sun, Athanasios V. Vasilakos
2021 arXiv   pre-print
In addition, a new DVO architecture, called deep direct sparse odometry (DDSO), is proposed to overcome the drawbacks of the previous direct sparse odometry (DSO) framework by embedding TrajNet.  ...  Traditional monocular direct visual odometry (DVO) is one of the most famous methods to estimate the ego-motion of robots and map environments from images simultaneously.  ...  Here, we focus on the combination of a geometry-based framework and deep learning-based pose or depth estimation. To improve the performance of geometry-based visual odometry, Tateno et al.  ... 
arXiv:1912.05101v3 fatcat:ooa4y7aoxbhf3a2mpeanw7iwda

Reducing drift in visual odometry by inferring sun direction using a Bayesian Convolutional Neural Network

Valentin Peretroukhin, Lee Clement, Jonathan Kelly
2017 2017 IEEE International Conference on Robotics and Automation (ICRA)  
We leverage recent advances in Bayesian Convolutional Neural Networks to train and implement a sun detection model that infers a three-dimensional sun direction vector from a single RGB image.  ...  We present a method to incorporate global orientation information from the sun into a visual odometry pipeline using only the existing image stream, where the sun is typically not visible.  ...  BCNNs rely on a connection between stochastic regularization (e.g. dropout, a widely adopted technique in deep learning) and approximate variational inference of a Bayesian Neural Network.  ... 
doi:10.1109/icra.2017.7989235 dblp:conf/icra/PeretroukhinCK17 fatcat:luwfbtob5fhcfka3njyerhx2ve

VLocNet++: Deep Multitask Learning for Semantic Visual Localization and Odometry

Noha Radwan, Abhinav Valada, Wolfram Burgard
2018 IEEE Robotics and Automation Letters  
While deep learning has enabled recent breakthroughs across a wide spectrum of scene understanding tasks, its applicability to state estimation tasks has been limited due to the direct formulation that  ...  In this work, we propose the VLocNet++ architecture that employs a multitask learning approach to exploit the inter-task relationship between learning semantics, regressing 6-DoF global pose and odometry  ...  While deep learning has enabled recent breakthroughs across a wide spectrum of scene understanding tasks, its applicability to state estimation tasks has been limited due to the direct formulation that  ... 
doi:10.1109/lra.2018.2869640 dblp:journals/ral/RadwanVB18 fatcat:vjrqodejefgonbqti7xtfvw64a

Sparse2Dense: From direct sparse odometry to dense 3D reconstruction [article]

Jiexiong Tang, John Folkesson, Patric Jensfelt
2019 arXiv   pre-print
In this paper, we proposed a new deep learning based dense monocular SLAM method.  ...  Compared to existing methods, the proposed framework constructs a dense 3D model via a sparse to dense mapping using learned surface normals.  ...  Monocular VO and SLAM Impressive progress has been made in visual odometry and SLAM methods. A common way to categorize different approaches is to use direct / indirect and dense / sparse.  ... 
arXiv:1903.09199v1 fatcat:tkkhtieguzfeldkccw7eptcnna

Improving Monocular Visual Odometry Using Learned Depth [article]

Libo Sun, Wei Yin, Enze Xie, Zhengrong Li, Changming Sun, Chunhua Shen
2022 arXiv   pre-print
Monocular visual odometry (VO) is an important task in robotics and computer vision.  ...  In this paper, we propose a framework to exploit monocular depth estimation for improving VO.  ...  SVO is a semi-direct monocular visual odometry approach, which is a combination of feature methods and direct methods.  ... 
arXiv:2204.01268v1 fatcat:wzjfyftcfvdgpc2i7s2dhbnpc4

Approaches, Challenges, and Applications for Deep Visual Odometry: Toward to Complicated and Emerging Areas [article]

Ke Wang, Sai Ma, Junlan Chen, Fan Ren
2020 arXiv   pre-print
Visual odometry (VO) is a prevalent way to deal with the relative localization problem, which is becoming increasingly mature and accurate, but it tends to be fragile under challenging environments.  ...  Therefore, this paper aims to gain a deep insight on how deep learning can profit and optimize the VO systems.  ...  approach to improve the accuracy and robustness of visual odometry.  ... 
arXiv:2009.02672v1 fatcat:zdnwt4lpmvbiromtpcxhxxpjxa

Learning Multiplicative Interactions with Bayesian Neural Networks for Visual-Inertial Odometry [article]

Kashmira Shinde, Jongseok Lee, Matthias Humt, Aydin Sezgin, Rudolph Triebel
2020 arXiv   pre-print
This paper presents an end-to-end multi-modal learning approach for monocular Visual-Inertial Odometry (VIO), which is specifically designed to exploit sensor complementarity in the light of sensor degradation  ...  Importantly, our work thereby provides an empirical evidence that learning multiplicative interactions can result in a powerful inductive bias for increased robustness to sensor failures.  ...  We consider Deep learning (DL) based approaches for visual localization. Amongst many, we review the most related work as follows.  ... 
arXiv:2007.07630v1 fatcat:pcilesijxbaunlw5jol4sg7h5i

Detection of road objects with small appearance in images for autonomous driving in various traffic situations using a deep learning based approach

Guofa Li, Heng Xie, Weiquan Yan, Yunlong Chang, Xingda Qu
2020 IEEE Access  
PROPOSED APPROACH To address the above problems, a deep learning method inspired by CenterNet was proposed, namely atrous spatial pyramid pooling (ASPP)-CenterNet.  ...  Section 3 illustrates the deep learning road object detection approach using atrous spatial pyramid pooling.  ... 
doi:10.1109/access.2020.3036620 fatcat:zjp6yxa245axjjn4yzlu47glr4

Self-Supervised Steering Angle Prediction for Vehicle Control Using Visual Odometry [article]

Qadeer Khan, Patrick Wenzel, Daniel Cremers
2021 arXiv   pre-print
In this paper, we demonstrate how a model can be trained to control a vehicle's trajectory using camera poses estimated through visual odometry methods in an entirely self-supervised fashion.  ...  Vision-based learning methods for self-driving cars have primarily used supervised approaches that require a large number of labels for training.  ...  An image is fed to a deep neural network which is trained in a self-supervised manner using camera poses obtained by visual odometry.  ... 
arXiv:2103.11204v1 fatcat:njzkakxhzjf4blesrs6xkwq3em

RP-VIO: Robust Plane-based Visual-Inertial Odometry for Dynamic Environments [article]

Karnik Ram, Chaitanya Kharyal, Sudarshan S. Harithas, K. Madhava Krishna
2021 arXiv   pre-print
a state-of-the-art monocular visual-inertial odometry system.  ...  We present RP-VIO, a monocular visual-inertial odometry system that leverages the simple geometry of these planes for improved robustness and accuracy in challenging dynamic environments.  ...  We focus on closely related visual-inertial odometry of existing approaches.  ... 
arXiv:2103.10400v2 fatcat:tlpg4v6hxrcw5cw45ta5wdp4bm

CKF-Based Visual Inertial Odometry for Long-Term Trajectory Operations

Trung Nguyen, George K. I. Mann, Andrew Vardy, Raymond G. Gosine
2020 Journal of Robotics  
The estimation error accumulation in the conventional visual inertial odometry (VIO) generally forbids accurate long-term operations.  ...  may not be feasible to be implemented in a low-cost robotic platform.  ...  It only affects the orientation and also requires considerable resources to train and execute the deep learning model of sun detection.  ... 
doi:10.1155/2020/7362952 fatcat:wvvnjtcjpfhrzbcu2n4bx3qpl4

Sensors and Sensor Fusion Methodologies for Indoor Odometry: A Review

Mengshen Yang, Xu Sun, Fuhua Jia, Adam Rushworth, Xin Dong, Sheng Zhang, Zaojun Fang, Guilin Yang, Bingjian Liu
2022 Polymers  
Therefore, this paper straightens the pathway of indoor odometry from principle to application. Finally, some future prospects are discussed.  ...  Modern sensors and algorithms endow moving robots with the capability to perceive their environment, and enable the deployment of novel localization schemes, such as odometry, or Simultaneous Localization  ...  of depth, pose, and uncertainty into direct visual odometry.  ... 
doi:10.3390/polym14102019 pmid:35631899 fatcat:nx4abgwuizbubgc376fxtpcqla

Stereo Visual Odometry and Semantics based Localization of Aerial Robots in Indoor Environments

Hriday Bavle, Stephan Manthe, Paloma de la Puente, Alejandro Rodriguez-Ramos, Carlos Sampedro, Pascual Campoy
2018 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
In this paper we propose a particle filter localization approach, based on stereo visual odometry (VO) and semantic information from indoor environments, for mini-aerial robots.  ...  We validate our approach in several real flight experiments where we compare it against ground truth and a state of the art visual SLAM approach.  ...  Stereo Visual Odometry Visual odometry can be performed either by a monocular or a stereo setup, although in monocular cases its difficult to estimate the true scale of the estimated trajectory and its  ... 
doi:10.1109/iros.2018.8593426 dblp:conf/iros/BavleMPRSC18 fatcat:d7km7glw4nf7tkk3zy7lr4r66y

IT-SVO: Improved Semi-Direct Monocular Visual Odometry Combined with JS Divergence in Restricted Mobile Devices

Chang Liu, Jin Zhao, Nianyi Sun, Qingrong Yang, Leilei Wang
2021 Sensors  
A more suitable methodology for SVO is that explores to improve the accuracy and robustness of mobile devices in unknown environments.  ...  In this paper, we combine information theory with classical visual odometry (SVO) and take Jensen-Shannon divergence (JS divergence) instead of Kullback-Leibler divergence (KL divergence) to estimate the  ...  We will then combine information theory to optimize and improve the VIO systems whose back-end is probability-based such as multi-state constraint Kalman filter (MSCKF).  ... 
doi:10.3390/s21062025 pmid:33809347 pmcid:PMC7998773 fatcat:r7bhzfj7bja5dkuvm2g2do5byy
« Previous Showing results 1 — 15 out of 353 results