Article
Engineering, Marine
Christopher Gotts, Benjamin Hall, Oliver Beaumont, Ziyang Chen, William Cleaver, James England, David White, Blair Thornton
Summary: This paper proposes an Autonomous Riser Inspection System (ARIS) for regular inspection of riser cables and data collection for future cable design improvement. Novel robotic methods for cable attachment, traversal, and inspection are described. The system enables autonomous navigation and position estimation, as well as damage identification and measurement of the cable.
Article
Computer Science, Theory & Methods
Rodrigo Henrique Cunha Palacios, Joao Paulo Scarabelo Bertoncini, Gabriel Henrique Oliveira Uliam, Marcio Mendonca, Lucas Botoni de Souza
Summary: The demand for mobile robotics applications has increased significantly in recent years, especially with the rise of industry 4.0 and autonomous robotics. Computer vision has emerged as an interesting alternative for controlling the movement of mobile robots, and various vision techniques have gained prominence. This research proposes the development of a control center using the global view technique for navigation and location of mobile robots in closed environments. Additionally, wireless communication (WiFi) between the exchange and the robots is being investigated. Promising results have been obtained in the initial stages of the project's development, with data collected from an autonomous robot compared to a human-guided robot. The A* and Dijkstra algorithms are employed for optimizing the trajectory between the origin and destination.
Review
Robotics
Vishnu Rajendran, Bappaditya Debnath, Sariah Mghames, Willow Mandil, Soran Parsa, Simon Parsons, Amir Ghalamzan-E.
Summary: This paper provides an overview of the current state-of-the-art in selective harvesting robots (SHRs) and their potential for addressing the challenges of global food production. It discusses the main components of SHRs, the challenges in developing SHR technologies, and the potential benefits of integrating artificial intelligence and soft robots. The paper also identifies open research questions in the field and highlights the need for further research and development efforts to advance SHR technologies.
JOURNAL OF FIELD ROBOTICS
(2023)
Article
Chemistry, Multidisciplinary
Feiyu Jia, Misha Afaq, Ben Ripka, Quamrul Huda, Rafiq Ahmad
Summary: This study focuses on the autonomous docking and recharging tasks of mobile robots in manufacturing environments. A deep learning model and Lidar sensor-based method for object detection and docking alignment are proposed. The developed method achieved an average accuracy of 95% in real-world scenarios. Fusion of vision-based and Lidar-based methods can improve the overall accuracy of docking alignment.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Marine
Ioannis Polymenis, Maryam Haroutunian, Rose Norman, David Trodden
Summary: Underwater vehicles have become more sophisticated due to advancements in underwater operations. This study utilizes recent advancements in deep learning to construct a bespoke dataset for underwater applications using generative adversarial networks.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Chemistry, Analytical
Waseem Akram, Alessandro Casavola, Nadir Kapetanovic, Nikola Miskovic
Summary: Aquaculture net pens inspection and monitoring are crucial for ensuring net stability and fish health in fish farms. Remotely operated vehicles (ROVs) provide a cost-effective and sophisticated solution for underwater fish net pens inspection due to their visual sensing and autonomy capabilities in challenging aquaculture environments. This paper presents an integration of an ROV with a visual servoing scheme for regular inspection and tracking of net pens, utilizing a vision-based positioning scheme and closed-loop control to traverse along the net plane for status inspection. Extensive experimental results have shown satisfactory performance, supplementing traditional aquaculture net pens inspection systems.
Article
Chemistry, Multidisciplinary
Anna Annusewicz-Mistal, Dawid Sebastian Pietrala, Pawel Andrzej Laski, Jaroslaw Zwierzchowski, Krzysztof Borkowski, Gabriel Bracha, Kamil Borycki, Szczepan Kostecki, Daniel Wlodarczyk
Summary: This article introduces a system for the autonomous operation of a manipulator, which can handle panels or grasp and move objects without human control. The system relies solely on a digital camera and uses markers to identify the position of the end-effector. Test results show that the system operates well under good lighting conditions. It is effective for activities that do not require high accuracy, but may not be suitable for high-precision tasks.
APPLIED SCIENCES-BASEL
(2023)
Article
Biochemistry & Molecular Biology
Baoguo Xu, Deping Liu, Muhui Xue, Minmin Miao, Cong Hu, Aiguo Song
Summary: In this study, a continuous shared control strategy combining continuous BCI and autonomous navigation was proposed for a mobile robot system. The system utilizes visual SLAM method to construct environmental maps and evaluates safe and reachable trajectories using brain-based shared control dynamic window approach. The results of training and online experiments demonstrated the feasibility and effectiveness of the proposed system in navigation tasks for brain-driven assistive mobile robots.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2023)
Article
Remote Sensing
Yago M. R. da Silva, Fabio A. A. Andrade, Lucas Sousa, Gabriel G. R. de Castro, Joao T. Dias, Guido Berger, Jose Lima, Milena F. F. Pinto
Summary: This research describes the potential of an autonomous unmanned aerial system (UAS) based on image processing technique and Convolutional Neural Network (CNN) strategy to navigate appropriately in complex unburied pipeline networks contributing to the monitoring procedure of the Oil & Gas Industry structures. The framework is assessed through simulations and real tests, showing robustness and feasibility in detecting pipes and following paths.
Article
Chemistry, Analytical
Povendhan Palanisamy, Rajesh Elara Mohan, Archana Semwal, Lee Ming Jun Melivin, Braulio Felix Gomez, Selvasundari Balakrishnan, Karthikeyan Elangovan, Balakrishnan Ramalingam, Dylan Ng Terntzer
Summary: This study introduces an AI-enabled robot-assisted framework for drain inspection and mapping, evaluated through deep learning and real-time trials. Results show that the robot's maneuverability was stable, with accurate mapping and localization in various drain types.
Article
Engineering, Civil
Peirong Wu, Airong Liu, Jiyang Fu, Xijun Ye, Yinghao Zhao
Summary: In this research, an improved YOLOv4 network with pruning technique and EvoNorm-S0 structure was proposed to better identify concrete cracks. Compared with other algorithms, this network shows superior performance in detecting concrete cracks.
ENGINEERING STRUCTURES
(2022)
Article
Automation & Control Systems
Nur Aira Abd Rahman, Khairul Salleh Mohamed Sahari, Nasri A. Hamid
Summary: This study proposes a practical solution to automatically localize and estimate the radiation intensity of hotspots within clutter in real-time using a mobile robot. By fusing measured data and taking into account the limitations of the detector, the method effectively identifies and prevents misclassifications of low radiation intensity hotspots. The results demonstrate the versatility and performance of the approach.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Chemistry, Multidisciplinary
Tongpo Zhang, Yunze Song, Zejian Kong, Tiantian Guo, Miguel Lopez-Benitez, Enggee Lim, Fei Ma, Limin Yu
Summary: This paper discusses the challenges of robot tracking under partial occlusion and compares the system performance of three recent DL models. A series of experiments are conducted to analyze the performance metrics under different scenarios and settings. Based on the metrics, a comparative metric P is devised to further compare the overall performance of the three DL models. The SSD model achieved the highest P score, outperforming the Faster RCNN and YOLOv5 models in both testing data sets.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Alvari Seppanen, Jari Vepsalainen, Risto Ojala, Kari Tammi
Summary: This article proposes semi-autonomous control strategies to assist in the teleoperation of mobile robots under unstable communication conditions. A short-term autonomous control system is used to assist in the semi-autonomous control strategies when teleoperation is compromised. Experimental results show that autonomous, delay-dependent, and control-dependent assist improves teleoperation compared to fully manual control.
JOURNAL OF INTELLIGENT & ROBOTIC SYSTEMS
(2022)
Article
Engineering, Electrical & Electronic
Zeba Khanam, Bilal Aslam, Sangeet Saha, Xiaojun Zhai, Shoaib Ehsan, Rustam Stolkin, Klaus McDonald-Maier
Summary: This paper examines the effects of gamma radiation on a robot vision sensor used for radiological inspection. Experimental results show that high dose rates of gamma radiation can lead to unreliable visual odometry, but the images can still be used for surveillance purposes.
IEEE SENSORS JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Guilherme Heim Weber, Hector Lise de Moura, Daniel Rodrigues Pipa, Cicero Martelli, Jean Carlos Cardozo da Silva, Marco Jose da Silva
Summary: This paper proposes a Sparse Inverse Chirp Z Transform approach designed for sparse time-domain signals representing localized fault locations in cables. Two algorithms, OMP and FISTA, are introduced for conversion, with verified performances in comparison with reference values for simulated and experimental data.
IEEE SENSORS JOURNAL
(2021)
Article
Automation & Control Systems
Raul de Queiroz Mendes, Eduardo Godinho Ribeiro, Nicolas dos Santos Rosa, Valdir Grassi
Summary: Inferring image depth is a fundamental inverse problem in Computer Vision, and Single Image Depth Estimation (SIDE) is highlighted for its low cost and robustness. State-of-the-art CNNs optimize the SIDE task for autonomous navigation, though supervised by sparse and noisy depth data from LiDAR scans at high computational cost.
ROBOTICS AND AUTONOMOUS SYSTEMS
(2021)
Article
Automation & Control Systems
Angelica Tiemi Mizuno Nakamura, Valdir Grassi Jr, Denis Fernando Wolf
Summary: This paper proposes a greedy approach to model multi-task learning as a multi-objective optimization problem, finding weighting coefficients for each task to balance the optimization of multiple loss functions. Experimental results show that enhancing instance segmentation by properly exploring depth information, as well as instance segmentation helping depth estimations, leads to better performance compared to single-task models.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2021)
Article
Engineering, Civil
Junior Anderson Rodrigues da Silva, Iago Pacheco Gomes, Denis Fernando Wolf, Valdir Grassi
Summary: This paper proposes a road network model based on clothoids for Autonomous Vehicles to autonomously navigate in traffic roads. Through piecewise linear continuous-curvature paths, the model takes into account the vehicle's compliance with traffic rules in urban scenarios, while also considering passengers' comfort parameters.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Gustavo A. Prudencio de Morais, Lucas Barbosa Marcos, Filipe Marques Barbosa, Bruno H. G. Barbosa, Marco Henrique Terra, Valdir Grassi
Summary: This study proposes a robust recursive controller designed via multiobjective optimization to overcome the challenges of system uncertainties, along with a local search method for multiobjective optimization problems. This method is applicable to any established multiobjective evolutionary algorithm.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Mechanical
Jefferson dos Santos Ambrosio, Andre Eugenio Lazzaretti, Daniel Rodrigues Pipa, Marco Jose da Silva
Summary: The paper introduces an objective approach for classifying flow patterns in two-phase flow using time series data, signal processing, and machine learning. Experimental results demonstrate high accuracy and F-score in real systems, showing efficiency and generalization of the proposed method.
FLOW MEASUREMENT AND INSTRUMENTATION
(2022)
Article
Chemistry, Analytical
Juliano Scholz Slongo, Jefferson Gund, Thiago Alberto Rigo Passarin, Daniel Rodrigues Pipa, Julio Endress Ramos, Lucia Valeria Arruda, Flavio Neves Junior
Summary: This article discusses the errors in flaw positioning and sizing that can occur when using ultrasonic inspection techniques at high working temperatures and proposes a mathematical tool for correcting focal laws to mitigate these errors. By considering acoustic anisotropy and temperature gradients, the tool significantly reduces flaw positioning errors.
Article
Computer Science, Artificial Intelligence
Angelica Tiemi Mizuno Nakamura, Valdir Grassi Jr, Denis Fernando Wolf
Summary: Multi-Task Learning is a learning paradigm that uses correlated tasks to improve performance generalization. We propose a method that takes into account the temporal behavior of gradients to adjust the importance of each task, addressing the problem of conflicting tasks in multi-task learning.
Article
Engineering, Electrical & Electronic
Frederico A. Jahnert, Guilherme H. Weber, Danilo F. Gomes, Marco J. da Silva, Daniel R. Pipa, Jean Carlos Cardozo da Silva, Cicero Martelli, Sergio T. Camargo Junior, Manoel F. Silva Junior, Jucelio T. Pereira, Carlos A. Bavastri
Summary: This article presents a novel method for measuring local dynamic strain of structures using distributed acoustic sensors (DASs). Experimental results show that using a serpentine configuration can significantly improve the accuracy of DAS measurements and increase the peak-to-noise ratio.
IEEE SENSORS JOURNAL
(2022)
Article
Chemistry, Analytical
Filipe C. A. Lins, Nicolas S. S. Rosa, Valdir Grassi Jr, Adelardo A. D. Medeiros, Pablo J. J. Alsina
Summary: Nowadays, state-of-the-art direct visual odometry (VO) methods primarily use points as features to estimate camera pose and reconstruct the environment. However, there are recent developments in monocular plane-based DSO techniques, with one using a learning-based plane estimator and the other only detecting planes in specific orientations. This article introduces the first Stereo Plane-based VO inspired by the DSO framework, using planes as features and dual quaternion as pose parameterization. Experimental results show comparable performance to the point-based Stereo DSO approach.
Proceedings Paper
Robotics
Gabriel Soares Gama, Nicolas dos Santos Rosa, Valdir Grassi Jr
Summary: This paper proposes adding a semantic segmentation decoder in a shared encoder architecture to improve the learning of semantic information in the feature extractor. This approach is more robust compared to using high-level semantic information alone, as it is intrinsically learned and not dependent on the quality of semantic prediction.
2022 LATIN AMERICAN ROBOTICS SYMPOSIUM (LARS), 2022 BRAZILIAN SYMPOSIUM ON ROBOTICS (SBR), AND 2022 WORKSHOP ON ROBOTICS IN EDUCATION (WRE)
(2022)
Article
Astronomy & Astrophysics
Alexandre J. T. S. Mello, Elder Oroski, Victor B. Frencl, Guido Agapito, Daniel R. Pipa
Summary: This paper investigates the robust control technique of pseudo open loop control for wide-field atmospheric turbulence compensation in multi-conjugated adaptive optics systems. Through system identification, open-loop and closed-loop frequency response analysis, a better tuning and characterization of the POLC system can be achieved.
JOURNAL OF ASTROPHYSICS AND ASTRONOMY
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Augusto R. Castro, Valdir Grassi Jr, Moacir A. Ponti
Summary: This study proposes a method for completing depth maps using low-cost depth-sensing devices. The method consists of a U-Net and a refinement module, and applies the Euclidean distance transform to increase the influence of missing pixels and reduce blur in predictions. The method outperforms state-of-the-art methods for completed depth maps in a benchmark dataset.
PROCEEDINGS OF THE 17TH INTERNATIONAL JOINT CONFERENCE ON COMPUTER VISION, IMAGING AND COMPUTER GRAPHICS THEORY AND APPLICATIONS (VISAPP), VOL 4
(2022)
Proceedings Paper
Robotics
Vitor A. S. Silva, Valdir Grassi Jr
Summary: This study combines Conditional Learning and Proximal Policy Optimization (PPO) to tackle the problem of turning at intersections and lane keeping in an end-to-end deep reinforcement learning (DRL) fashion. Three PPO sub-policies were trained for turning and lane following, showing good performance in experiments. Properly choosing image transformations can improve sample efficiency and generalization capability.
2021 LATIN AMERICAN ROBOTICS SYMPOSIUM / 2021 BRAZILIAN SYMPOSIUM ON ROBOTICS / 2021 WORKSHOP OF ROBOTICS IN EDUCATION (LARS-SBR-WRE 2021)
(2021)
Proceedings Paper
Robotics
Anderson Carlos dos Santos, Valdir Grassi Junior
Summary: This article introduces a model for predicting pedestrian future trajectories, using key points of pedestrians as inputs and exploring the benefits of automatically extracting features. The final approach reduces the mean squared error of pedestrian future trajectories without the need for human annotators.
2021 LATIN AMERICAN ROBOTICS SYMPOSIUM / 2021 BRAZILIAN SYMPOSIUM ON ROBOTICS / 2021 WORKSHOP OF ROBOTICS IN EDUCATION (LARS-SBR-WRE 2021)
(2021)