Review
Automation & Control Systems
Jun Cheng, Liyan Zhang, Qihong Chen, Xinrong Hu, Jingcao Cai
Summary: This article discusses the demand for visual SLAM technology in autonomous driving vehicles, focusing on the typical structure, latest research findings, and future trends of visual SLAM.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Information Systems
Vijay John, Seiichi Mita
Summary: Object detection is an important perception task in autonomous driving, and a deep learning framework for effective sensor fusion of the visible camera with complementary sensors has been proposed in this study. The feature-level sensor fusion technique, utilizing skip connection, is better than baseline early and late fusion frameworks, as shown by the results obtained from public datasets.
Article
Engineering, Electrical & Electronic
Kelvin Wong, Yanlei Gu, Shunsuke Kamijo
Summary: This article provides a comprehensive overview of the production and usage of maps for autonomous driving, highlighting the critical role maps play in enabling safe and successful autonomous driving. While there are various challenges in mapping that need to be addressed, steps like developing efficient infrastructure, setting minimum data quality requirements, and establishing international mapping standards can help move towards realizing autonomous driving technology. The article concludes with 11 open research challenges for the field.
IEEE INTELLIGENT TRANSPORTATION SYSTEMS MAGAZINE
(2021)
Article
Chemistry, Analytical
Hengyuan Zhang, Shashank Venkatramani, David Paz, Qinru Li, Hao Xiang, Henrik I. Christensen
Summary: Statistical learning techniques and increased computational power have enabled the development of self-driving car technology. However, the high cost of scaling and maintaining high-definition (HD) maps has been a limiting factor. In response, we present an approach that combines pre-built point cloud map data with images to identify static landmarks accurately. Our pipeline uses semantic segmentation of 2D images, associates semantic labels with points in point cloud maps, and generates a probabilistic bird's-eye view semantic map. The approach has been tested in an urban area and can be extended to automatic generation of HD maps. The software pipeline is implemented in ROS and made available.
Article
Chemistry, Analytical
Xavier Dauptain, Aboubakar Kone, Damien Grolleau, Veronique Cerezo, Manuela Gennesseaux, Minh-Tan Do
Summary: This paper presents a high level, compact, scalable, and long autonomy perception and localization system for autonomous driving applications. The system consists of a high resolution lidar, a stereo global shutter camera, an inertial navigation system, a time server, and an embedded computer. It incorporates two perception algorithms (RBNN detection, DCNN detection) and one localization algorithm (lidar-based localization) to provide real-time advanced information such as object detection and localization in challenging environments. A dataset of 10,000 annotated lidar frames is used to train and evaluate the algorithms. The performance of the algorithms is competitive with the state-of-the-art and their processing time is compatible with real-time autonomous driving applications. By offering accurate advanced outputs, this system aims to simplify the work of researchers and engineers in planning and control modules, contributing to the democratization of access to autonomous vehicle research platforms.
Article
Computer Science, Artificial Intelligence
Kashyap Chitta, Aditya Prakash, Bernhard Jaeger, Zehao Yu, Katrin Renz, Andreas Geiger
Summary: In this study, a method called TransFuser is proposed to integrate representations from both images and LiDAR. By using self-attention mechanism to fuse feature maps at different resolutions, the method achieves better performance in complex driving scenarios.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Engineering, Civil
Eduardo Arnold, Mehrdad Dianati, Robert de Temple, Saber Fallah
Summary: This article investigates two schemes for cooperative 3D object detection - early fusion and late fusion, and evaluates their performance in complex driving scenarios. The results show that early fusion outperforms late fusion, demonstrating the advantages of cooperative perception over single-point sensing.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Civil
Muhammad Mobaidul Islam, Abdullah Al Redwan Newaz, Ali Karimoddini
Summary: This paper proposes a novel fusion framework that combines asymmetric inferences from object detectors and semantic segmentation networks for jointly detecting multiple pedestrians. By introducing a consensus-based scoring method to fuse pair-wise pixel-relevant information, the final confidence scores are boosted with low runtime overhead through parallel implementation.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Environmental Sciences
Min-Su Kang, Jae-Hoon Ahn, Ji-Ung Im, Jong-Hoon Won
Summary: This paper introduces a cooperative positioning technique for autonomous vehicles, which utilizes lidar sensor and vehicle-to-everything communication to address problems in vehicle localization in GNSS-denied situations.
Article
Engineering, Electrical & Electronic
Yasser H. Khalil, Hussein T. Mouftah
Summary: Researchers propose enhancing urban autonomous driving using multi-modal fusion with latent deep reinforcement learning. The method extracts and fuses images from multiple sensors to predict vehicle perception and motion, and then trains a driving policy using latent deep reinforcement learning to ensure safety, efficiency, and comfort. Experimental results show that the proposed method outperforms other existing models.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2023)
Article
Robotics
LeiChen Wang, Bastian Goldluecke
Summary: The method proposed in this study effectively addresses the issue of significant performance degradation in detecting objects beyond 50 meters by introducing a new key point sampling algorithm and dynamic continuous occupancy heatmap, achieving superior performance in the far range while maintaining comparable performance in the near range.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Information Systems
Xinhu Zheng, Yuru Li, Dongliang Duan, Liuqing Yang, Chen Chen, Xiang Cheng
Summary: In autonomous driving, environment perception is crucial, but existing methods suffer from measurement errors and blind spots. To address these issues, a multivehicle and multisensor cooperative perception method is proposed, which constructs an occupancy grid map to provide comprehensive and raw environmental information. This method expands the perception range and captures sensor data uncertainty more effectively.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Review
Environmental Sciences
Shuran Zheng, Jinling Wang, Chris Rizos, Weidong Ding, Ahmed El-Mowafy
Summary: The Simultaneous Localization and Mapping (SLAM) technique has made significant progress in recent decades and has attracted considerable attention in the autonomous driving community. This study provides an overview of different SLAM implementation approaches and discusses the applications, challenges, and solutions in the context of autonomous driving. It also presents a real-world road test showcasing a multi-sensor-based SLAM procedure for autonomous driving.
Article
Engineering, Civil
Mehmood Nawaz, Jeff Kai-Tai Tang, Khadija Bibi, Shunli Xiao, Ho-Pui Ho, Wu Yuan
Summary: This paper outlines the vital role of sensor fusion in intelligent transportation systems, discussing the capabilities, impacts, technological challenges, and future solutions for the automotive industry.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Engineering, Civil
Zakaria Bairi, Olfa Ben-Ahmed, Abdenour Amamra, Abbas Bradai, Kadda Beghdad Bey
Summary: This paper proposes a deep neural network for Compressed Sensing (CS) based image reconstruction to improve visual prediction tasks in autonomous driving cars. The method outperforms existing approaches in terms of both image reconstruction quality and processing time, and the quality of the reconstructed images is evaluated through semantic scene segmentation.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Abtin Djavadifar, John Brandon Graham-Knight, Marian Korber, Patricia Lasserre, Homayoun Najjaran
Summary: The paper discusses the evaluation of four deep convolutional neural network models for the detection of fiber composite material boundaries and defects in the aviation industry. While good detection accuracy is achieved for gripper and fabric based on IoU scores, wrinkle detection shows lower accuracy due to geometrical ambiguities. The model is found to outperform a human operator in certain aspects and new approaches are introduced for wrinkle detection.
JOURNAL OF INTELLIGENT MANUFACTURING
(2022)
Review
Engineering, Environmental
George Luka, Ehsan Samiei, Nishat Tasnim, Arash Dalili, Homayoun Najjaran, Mina Hoorfar
Summary: Cryptosporidium, a critical waterborne protozoan pathogen, has been a major cause of death and serious illnesses worldwide, with high costs for detection and treatment. Current detection methods have limitations and there is a need for improving and developing low-cost, portable, and rapid detection tools for applications in the water quality industry. Recent advances in nanotechnology, biosensing, and microfluidics have facilitated the development of sophisticated tools for detecting Cryptosporidium.
JOURNAL OF HAZARDOUS MATERIALS
(2022)
Review
Engineering, Industrial
Ardeshir Shojaeinasab, Todd Charter, Masoud Jalayer, Maziyar Khadivi, Oluwaseyi Ogunfowora, Nirav Raiyani, Marjan Yaghoubi, Homayoun Najjaran
Summary: This paper systematically reviews the research and implementations in Manufacturing Execution Systems (MES) to identify promising research topics and proposes a conceptual framework for Intelligent MES (IMES).
JOURNAL OF MANUFACTURING SYSTEMS
(2022)
Review
Computer Science, Artificial Intelligence
Aria Salari, Abtin Djavadifar, Xiangrui Liu, Homayoun Najjaran
Summary: This article provides a detailed analysis of datasets in the highly investigated field of object recognition, summarizing statistical data and descriptions of over 160 datasets. It also introduces object recognition benchmarks and competitions, along with commonly adopted evaluation metrics in the computer vision community.
Article
Computer Science, Interdisciplinary Applications
Debasmita Mukherjee, Kashish Gupta, Li Hsin Chang, Homayoun Najjaran
Summary: In response to increased global competition, manufacturers are required to be more flexible in meeting customer demands, leading to the introduction of human operators and robots for their respective strengths, with a growing interest in shared human-robot workspace. Research in industrial human-robot collaboration focuses on human-robot safety, collaboration modes, and robot autonomy and adaptability.
ROBOTICS AND COMPUTER-INTEGRATED MANUFACTURING
(2022)
Article
Multidisciplinary Sciences
George S. Luka, Homayoun Najjaran, Mina Hoorfar
Summary: This study developed a simple, easy to fabricate, and cost-effective on-chip electrochemical biosensor for sensitive and label-free detection of Cryptosporidium oocysts in water samples. The sensor showed high sensitivity and specificity, making it a promising platform for fast, real-time, inexpensive, and label-free sensing for early warning and on-site detection.
SCIENTIFIC REPORTS
(2022)
Article
Automation & Control Systems
Zengjie Zhang, Dirk Wollherr, Homayoun Najjaran
Summary: This article presents a novel force-sensor-less method for estimating external forces in second-order robotic systems. The method utilizes an integral sliding mode observer (ISMO) as a second-order differentiator for position measurement. The ISMO allows for estimation of system states and disturbance without explicit force and velocity measurements. The method is evaluated through numerical simulation and compared with a conventional sliding mode observer (SMO), demonstrating its superior performance.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2022)
Article
Chemistry, Analytical
Maryam Ahang, Masoud Jalayer, Ardeshir Shojaeinasab, Oluwaseyi Ogunfowora, Todd Charter, Homayoun Najjaran
Summary: Bearings are essential components of rotating machines, but they are prone to unexpected faults. This study focuses on bearing fault diagnosis and condition monitoring to reduce operational costs and downtime. The proposed algorithm based on CGANs is able to generate fault data from normal data, improving fault diagnosis tools and optimizing operational performance and safety.
Article
Robotics
Debasmita Mukherjee, Kashish Gupta, Homayoun Najjaran
Summary: This paper provides a critical analysis of human-robot communication in industrial settings through the lens of Complexity Theory. It identifies research gaps in the field and suggests that utilizing natural communication models based on Complexity Theory can improve the accuracy and naturalness of human-robot communication.
FRONTIERS IN ROBOTICS AND AI
(2022)
Article
Computer Science, Artificial Intelligence
John Brandon Graham-Knight, Corey Bond, Homayoun Najjaran, Yves Lucet, Patricia Lasserre
Summary: This paper proposes finding network architectures that achieve similar performance while promoting diversity for ensembling. Prediction performance and diversity of various network sizes and activation functions applied to semantic segmentation are explained. The study shows that performance and diversity can be predicted from neural network architecture using explainable boosting machines. A grid search of 144 models is performed, and many of the models exhibit no significant difference in mean performance within a 95% confidence interval. It is found that the best performing models have varied network architecture parameters and diversity between models can be achieved by varying network size. Moreover, homogeneous network sizes generally show positive correlation in predictions and larger models tend to converge to similar solutions.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Robotics
Zengjie Zhang, Ram Dershan, Amir M. Soufi Enayati, Marjan Yaghoubi, Dean Richert, Homayoun Najjaran
Summary: Simulation is an efficient and safe evaluation solution for industrial automation, allowing software to be tested before being deployed in real systems. However, only high-fidelity simulation environments that accurately recreate the behavior of real systems can ensure a successful transition from simulation to reality. Existing industrial simulation tools often lack the ability to model system dynamics, resulting in unrealistic representations.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Automation & Control Systems
Yussuf Reza Esmaeili, Brett Cosco, Homayoun Najjaran
Summary: The presence of leakages in composite manufacturing vacuum bag layups is a concern for product quality. Traditional numerical methods often fail to estimate the leakage location accurately due to model inadequacies, large vacuum bags, complex geometries, and port configurations. Generating large datasets for machine learning is time-consuming and labor-intensive. A new analogy between vacuum bag assemblies and electrical circuits has been proposed for flowrate data generation, which has shown promising experimental validation and can be an accurate analogue for the real setup. Machine learning models have also been trained with a high validation accuracy of 94%, providing suitable predictions away from the boundaries.
INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY
(2023)
Review
Robotics
Mehran Ghafarian Tamizi, Marjan Yaghoubi, Homayoun Najjaran
Summary: Motion planning is crucial for successful robotic performance, and learning-based methods have gained attention for their ability to handle complex issues. This research provides an overview of recent developments in motion planning for manipulator robotics systems and explores learning-based methods to address the limitations of classical approaches.
INTERNATIONAL JOURNAL OF INTELLIGENT ROBOTICS AND APPLICATIONS
(2023)
Proceedings Paper
Automation & Control Systems
Kashish Gupta, Homayoun Najjaran
Summary: This paper presents a novel concept called EASE (Exploitation of Abstract Symmetry of Environments) that aims to improve the sample efficiency of traditional reinforcement learning algorithms by exploiting abstract spatial symmetry in complex environments. The concept is exemplified through three different settings and a novel algorithm is proposed for each setting.
2022 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2022)
(2022)
Proceedings Paper
Computer Science, Cybernetics
Debasmita Mukherjee, Kashish Gupta, Homayoun Najjaran
Summary: Cohesive human-robot collaboration requires natural communication between intelligent robots and humans. This study proposes an AI-powered multimodal fusion architecture to achieve more natural communication by dealing with uncertainty. The architecture can be applied in various human-machine communication scenarios.
2022 31ST IEEE INTERNATIONAL CONFERENCE ON ROBOT AND HUMAN INTERACTIVE COMMUNICATION (IEEE RO-MAN 2022)
(2022)