Article
Engineering, Electrical & Electronic
Huawei Hu, Kui Li, Wenwei Liang, Qihang Li, Ziyu Xie
Summary: This article proposes a method to improve the navigation accuracy of land vehicles by using kilometer signs to correct the errors of the inertial navigation system and odometer integration system. Absolute global positions of kilometer signs are obtained through image recognition and data matching, and vehicle visual positioning results are obtained using a monocular camera after compensating for the lever arm between the kilometer sign and INS. The vehicle positions from the kilometer signs are fused by a filter, and two Kalman filters are established to integrate the information from visual positioning, INS, and odometer. Simulation and vehicle experiment results demonstrate the effectiveness of the proposed method.
IEEE SENSORS JOURNAL
(2023)
Article
Automation & Control Systems
Marsela Car, Barbara Arbanas Ferreira, Jelena Vuletic, Matko Orsag
Summary: This article proposes a design for a farm called SPECULARIA, which is a heterogeneous robotic system that can address farming challenges and replace large farm machinery. The system uses mobile robots to move plants grown in containers and ensures optimal care and growth conditions for each plant. It utilizes compliance control algorithms combined with artificial intelligence to create a functional model of the plant and program different plant treatments. The proposed setup has been benchmarked and proven to be feasible.
IEEE ROBOTICS & AUTOMATION MAGAZINE
(2023)
Article
Robotics
Zhong Wang, Ying Shen, Lin Zhang, Shaoming Zhang, Yicong Zhou
Summary: This letter introduces a path planning and tracking method for carlike robots in coverage navigation, solving the challenges of stable path tracking and high coverage rate by introducing reference line constraints, heuristic initialization strategy, and cubic parametric path smoother. The proposed approach significantly reduces deviation off the reference line and achieves high success rate even in complex environments. The path tracker designed in this study ensures stable and high-precision tracking for multi-segment paths.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Artificial Intelligence
Xiang Wu, Zejia Zheng, Juyang Weng
Summary: Machine learning methods face challenges in achieving general-purpose learning due to complex situations in different tasks. The proposed general-purpose learning framework DN-2 autonomously generates internal hierarchy to form flexible representations, successfully learning different tasks in experiments. DN-2 is believed to be making progress towards fully autonomous learning.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Hai-Tao Zhang, Bin-Bin Hu, Zhecheng Xu, Zhi Cai, Bin Liu, Xudong Wang, Tao Geng, Sheng Zhong, Jin Zhao
Summary: This article presents a novel visual navigation and landing control system for an unmanned aerial vehicle to land on a moving autonomous surface vehicle. Experimental results validate the effectiveness of the proposed method.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Robotics
Yingqiang Wang, Ruoyu Hu, S. H. Huang, Zhikun Wang, Peizhou Du, Wencheng Yang, Ying Chen
Summary: Underwater positioning is crucial for navigation and geo-referencing of AUVs. Traditional methods like GPS are not feasible in underwater environments, leading to the importance of acoustic methods like USBL positioning systems. However, the high cost and complexity of classical USBL systems hinder their widespread adoption, which is addressed by the piUSBL positioning method.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Jing Dong, Xingyu Ren, Songlai Han, Shilin Luo
Summary: This paper proposes an approach to aid autonomous navigation without external GNSS aiding using vision positioning carried by a UAV. The method significantly improves navigation performance through data fusion and image matching.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2022)
Article
Telecommunications
Xingxing Li, Baoshan Song, Zhiheng Shen, Yuxuan Zhou, Hongbo Lyu, Zeyang Qin
Summary: In this paper, a new precise localization framework for autonomous robot systems based on inter-vehicle GNSS information fusion (IGIF) is presented. The framework incorporates a hybrid GNSS filter (HGF) and a consistency checking (CC) method to improve system reliability. The main contributions of this method are the fusion of ionosphere-free (IF) and double differenced (DD) combinations to enhance global positioning accuracy, as well as the two-level probabilistic detection of erroneous inter-vehicle measurements. Real-world experiments demonstrate that the proposed method achieves more consistent, continuous, and accurate position estimation compared to existing approaches while requiring low communication bandwidth.
IEEE COMMUNICATIONS LETTERS
(2023)
Article
Engineering, Marine
Xin Zhang, Bo He, Pengcheng Mu, Di Zhang
Summary: This paper proposes a Hybrid Model (HM) navigation methodology for Autonomous Underwater Vehicles (AUVs), which includes an error correction sub-model and a state regression sub-model. The HM algorithm achieves superior navigation accuracy and good fault tolerance capability.
Article
Robotics
Paul B. Reverdy, Vasileios Vasilopoulos, Daniel E. Koditschek
Summary: This article demonstrates a reactive sensorimotor architecture driven by motivation dynamics for mobile robots, which can achieve complex tasks by composing low-level control vector fields. The architecture shows strong robustness and can adapt to various disturbance situations.
IEEE TRANSACTIONS ON ROBOTICS
(2021)
Article
Robotics
Wei Zhu, Mitsuhiro Hayashibe
Summary: This article presents a Dreamer-based motion planner for collision-free navigation in diverse pedestrian scenarios. The method completely learns from zero experience through model-based deep reinforcement learning. Experimental results demonstrate that the motion planner outperforms existing methods and can be physically implemented in the real world.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Engineering, Civil
Lu Xiong, Rong Kang, Junqiao Zhao, Peizhi Zhang, Mingyu Xu, Ran Ju, Chen Ye, Tiantian Feng
Summary: This paper proposes a vehicle dynamics and intermittent GNSS-aided visual-inertial state estimator to address autonomous vehicle localization, achieving robust and accurate performance under various GNSS states.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Information Systems
Oualid Doukhi, Deok Jin Lee
Summary: This paper presents a novel approach for enabling a micro-aerial vehicle system equipped with sensors to autonomously navigate and explore unknown environments, achieving collision-free flights in the real world solely through training on a 3D simulator. The method does not require expert demonstrations, 3D mapping, or path planning.
Article
Robotics
Timothy L. Molloy, Tobias Fischer, Michael Milford, Girish N. Nair
Summary: The Bayesian Selective Fusion approach actively selects and fuses informative reference images, avoiding counterproductive fusion of all images and considering varying uncertainty to provide an effective VPR solution. It outperforms alternative fusion approaches and state-of-the-art techniques, making it well-suited for long-term robot autonomy in dynamic visual environments.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Chemistry, Multidisciplinary
Anas Al Tarabsheh, Maha Yaghi, AbdulRehman Younis, Razib Sarker, Sherif Moussa, Yazeed Eldigair, Hassan Hajjdiab, Ayman El-Baz, Mohammed Ghazal
Summary: This paper introduces an autonomous service robot for handling multiple teaching assistant duties in the educational field, which can be used to promote the transition to contactless learning activities during pandemics. The research results show that the robot can successfully navigate the environment to proctor exams and scan papers in educational activities.
APPLIED SCIENCES-BASEL
(2021)
Article
Engineering, Civil
Pierluigi Vito Amadori, Tobias Fischer, Yiannis Demiris
Summary: HammerDrive is a novel architecture for task-aware visual attention prediction in driving, able to reliably infer the driver's current focus of attention in real-time. The results show that HammerDrive outperforms a comparable state-of-the-art deep learning model in tracking driving maneuvers with significant improvement.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Robotics
Ahmad Khaliq, Michael Milford, Sourav Garg
Summary: In this paper, the authors propose an improved NetVLAD representation learning method, which utilizes a low-resolution image pyramid encoding to obtain richer place representations. The resulting multi-resolution feature pyramid can be easily aggregated using the VLAD algorithm, eliminating the need for concatenating or summing multiple patches. Experimental results demonstrate that the proposed method achieves state-of-the-art performance in global descriptor-based retrieval.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Connor Malone, Sourav Garg, Ming Xu, Thierry Peynot, Michael Milford
Summary: This research improves road segmentation based on similar places by eliminating limitations of traditional methods. By using Visual Place Recognition to find similar places and integrating a novel segmentation quality metric, the system achieves state-of-the-art road segmentation performance in multiple challenging scenarios, without the need for prior training or access to the same geographical locations.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Automation & Control Systems
Tobias Fischer, Wolf Vollprecht, Silvio Traversaro, Sean Yen, Carlos Herrero, Michael Milford
IEEE ROBOTICS & AUTOMATION MAGAZINE
(2022)
Article
Robotics
Bruno Ferrarini, Michael J. Milford, Klaus D. McDonald-Maier, Shoaib Ehsan
Summary: This article proposes a new class of highly compact models for visual place recognition (VPR) on resource-constrained platforms. By decreasing the precision of model parameters, reducing network depth, and decreasing the number of neurons in the classifier stage, the proposed models significantly reduce memory requirements and computational effort while maintaining state-of-the-art VPR performance.
IEEE TRANSACTIONS ON ROBOTICS
(2022)
Article
Robotics
Scarlett Raine, Ross Marchant, Brano Kusy, Frederic Maire, Tobias Fischer
Summary: Monitoring coral reefs using underwater vehicles and automated semantic segmentation can improve the range and availability of marine surveys. This study proposes a point label aware method for generating single-species segments in coral images, outperforming previous approaches and reducing computation time. The trained DeepLabv3+ model achieves impressive results in semantic segmentation on the UCSD Mosaics and Eilat datasets.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Communication
Mike Milford
Summary: In 2021, the creation and subsequent demise of the Euro Super League highlighted the power of collective rhetoric, particularly the use of kategoria as a form of accusation against the league's owners. The supporters of the six English Premier League clubs positioned themselves as moral authorities, criticizing the owners for their greed and disconnect from the spirit of English soccer. This case demonstrates the intersection of community, morality, and simplified moral issues in contemporary sports.
COMMUNICATION & SPORT
(2023)
Article
Hospitality, Leisure, Sport & Tourism
Mike Milford, Taylor J. Hendrickson
Summary: This research examines the role of Roger Bannister as a representative figure of the British identity and how parliament used him to shape and manage the country's image. The findings reveal the idealization of Bannister as a symbol of Britain's excellence and emphasize the need for careful management of his status. Athletes like Bannister are seen as instrumental in serving national purposes of identity construction and representation.
Article
Robotics
Tobias Fischer, Michael Milford
Summary: Event cameras are of interest due to their characteristics such as high dynamic range, low latency, minimal motion blur, and energy efficiency. This letter explores the distinctiveness of event streams from a small subset of pixels for visual place recognition, demonstrating effectiveness in matching query observations to reference places.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Stephen Hausler, Ming Xu, Sourav Garg, Punarjay Chakravarty, Shubham Shrivastava, Ankit Vora, Michael Milford
Summary: 6-DoF visual localization systems use 3D geometry to estimate camera pose accurately. By using hierarchical pipelines and learned 2D feature extractors, scalability and performance can be improved. This study investigates the usage of place-specific configurations to enhance worst-case localization performance, and demonstrates its effectiveness on the Ford AV benchmark dataset.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Robotics
Krishan Rana, Vibhavari Dasagi, Jesse Haviland, Ben Talbot, Michael Milford, Niko Sunderhauf
Summary: BCF is a hybrid control strategy that combines traditional hand-crafted controllers with model-free deep reinforcement learning. By fusing uncertainty-aware distributional outputs, BCF arbitrates control between the two systems and achieves better performance. It accelerates learning and ensures safe exploration and deployment.
INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Samuel Wilson, Tobias Fischer, Niko Sunderhauf, Feras Dayoub
Summary: We apply ideas from Hyperdimensional Computing to improve Out-of-Distribution (OOD) detection. Instead of using a single layer, our method uses semi-orthogonal projection matrices to project feature maps from multiple layers into a shared vector space, creating expressive descriptor vectors for all classes. By calculating cosine similarity, our method consistently identifies OOD samples with competitive performance and high efficiency. Our method can be combined with existing OOD detectors for further improvement.
2023 IEEE/CVF WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV)
(2023)
Article
Automation & Control Systems
Tobias Fischer, Wolf Vollprecht, Silvio Traversaro, Sean Yen, Carlos Herrero, Michael Milford
IEEE ROBOTICS & AUTOMATION MAGAZINE
(2022)