Article
Robotics
Yisoo Lee, Hosang Lee, Jinoh Lee, Jaeheung Park
Summary: The study develops an event-based finite-state machine (E-FSM) to enhance reactivity of biped robots to unforeseen disturbances for robust walking. The proposed control method ensures stable locomotion under external disturbances, as validated through experiments.
IEEE TRANSACTIONS ON ROBOTICS
(2022)
Article
Robotics
Nelson Rosa, Kevin M. Lynch
Summary: This article introduces a topological method for generating families of walking gaits for underactuated biped walkers, utilizing implicitly defined feasible periodic gaits within a state-time-control space. Equilibria are used as reliable templates for constructing gait families on several 2-D and 3-D biped walkers.
IEEE TRANSACTIONS ON ROBOTICS
(2022)
Article
Robotics
Guangrong Chen, Ningze Wei, HuaFeng Lu, Lei Yan, Jin Li
Summary: To improve the locomotion performance of legged robots, the swing leg retraction (SLR) technique in a hydraulic biped robot is investigated. The influence of SLR on the robot's locomotion performance is analyzed through theory, simulations, and experiments, considering energy loss/efficiency, friction/slipping, and impact/compliance. The results show that the SLR technique reduces impact force, improves locomotion stability, and allows for compliance controller improvement. This research provides valuable insights for locomotion control of hydraulic legged robots.
JOURNAL OF FIELD ROBOTICS
(2023)
Article
Robotics
Kim-Ngoc-Khanh Nguyen, Yuta Kojio, Shintaro Noda, Fumihito Sugai, Kunio Kojima, Yohei Kakiuchi, Kei Okada, Masayuki Inaba
Summary: The study proposed a method to optimize joint trajectories considering time and robot orientation variables, using evolutionary search in a dynamic simulator to achieve fall recovery for biped robots. By combining fall recovery feature with fall detector and fall-damage-reduction motion, a biped robot platform was successfully developed.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Computer Science, Information Systems
Fahad Raza, Wei Zhu, Mitsuhiro Hayashibe
Summary: The self-balancing wheel-legged robot has higher maneuverability and mobility, attracting interest from academia and the commercial sector. This paper focuses on using active arm control to improve the balance stability and robustness of the robot, providing important insights for future applications in real-world environments with human-robot interactions.
Article
Engineering, Biomedical
Tiange Zhang, David J. Braun
Summary: This study proposes a human-driven robot exoskeleton system that can assist humans in accelerating and maintaining high walking speeds, especially when carrying heavy loads. The exoskeleton system utilizes mechanically adaptive spring limbs that emulate the load-bearing mechanics of a bicycle, thus enhancing human weight-bearing and fast-walking abilities.
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
John Grezmak, Nicole Graf, Alexander Behr, Kathryn Daltorio
Summary: A novel approach for terrain classification for legged robots is presented, using low cost Hall effect magnetometers to gather terrain information. The method is demonstrated to be highly effective in a beach environment, achieving an accuracy of up to 99.3%. Combining information from different sensor modalities can lead to even higher accuracy in classification.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Mechanical
Linqi Ye, Xueqian Wang, Houde Liu, Bin Liang, Bo Yuan
Summary: This paper investigates how to walk faster for two simple 2D walking models. Open-loop analysis is conducted and the concept of acceleration factor is proposed. It is found that the acceleration factor has a fixed correlation with the velocity transition trend, independent of the step length. Based on this, walking controllers are designed and closed-loop simulations are performed to achieve faster walking speeds.
NONLINEAR DYNAMICS
(2023)
Article
Engineering, Electrical & Electronic
Luis Almeida, Vitor Santos, Joao P. Ferreira
Summary: This study presents a technique for real-time classification of floor types based on contact force data and robot inertial sensor information, achieving a classification accuracy of over 92%. Through testing different learning models and tuning parameters, a good mapping between inputs and targets is achieved, demonstrating the suitability of the proposed approach for the multi-classification problem addressed.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
Peter Billeschou, Carsten Albertsen, Jorgen Christian Larsen, Poramate Manoonpong
Summary: Ensuring feasibility and reliability in measuring forces and torques in legged robots is a challenging task. Different technologies can be used to develop sensors with multiple axes, high accuracy, and durability. Strain gauges, while high in accuracy, have design challenges such as large housing structures and susceptibility to error. A new design for a low-cost, compact, and hermetically sealed three-axis force/torque sensor is proposed in this study, demonstrating error reduction and accurate measuring capabilities in all angles.
IEEE SENSORS JOURNAL
(2021)
Article
Engineering, Multidisciplinary
Henry Chang, Justin Chang, Glenna Clifton, Nick Gravish
Summary: This study investigated a passive method for overcoming swing-collisions in legged locomotion by implementing virtual compliance control in a robot leg. The virtual compliance methods improved the successful negotiation of step obstacles by over 70%, highlighting the importance of limb compliance in enhancing walking performance in naturalistic environments.
BIOINSPIRATION & BIOMIMETICS
(2021)
Article
Engineering, Electrical & Electronic
Tzuu-Hseng S. Li, Ping-Huan Kuo, Chuan-Han Cheng, Chia-Ching Hung, Po-Chien Luan, Chien-Hsin Chang
Summary: This study proposes a real-time sequential sensor fusion based gait pattern controller for a bipedal humanoid robot, utilizing LSTM to build a feedback system for correcting walking parameters. The methodology is tested in simulation and on a real robot, demonstrating improved performance and self-adjustment capability.
IEEE SENSORS JOURNAL
(2021)
Article
Automation & Control Systems
Ping Sun, Rui Shan, Shuoyu Wang
Summary: A rehabilitation gait training robot has been developed, which can directly switch between passive and active training during walking, improving the intelligence and security of the robot and enhancing the effectiveness of rehabilitation training.
IEEE ROBOTICS & AUTOMATION MAGAZINE
(2023)
Article
Engineering, Multidisciplinary
Max Austin, Ashley Chase, Brian Van Stratum, Jonathan E. Clark
Summary: This study investigates multi-modal limb locomotion and develops a limb aquatic-scansorial multi-modal robot.
BIOINSPIRATION & BIOMIMETICS
(2023)
Article
Computer Science, Information Systems
Petr Cizek, Martin Zoula, Jan Faigl
Summary: This paper focuses on the characteristics of six-legged walking robots with statically-stable gaits, proposing a novel construction to improve their motion capabilities, speed, reliability, and endurance. Through experimental verification, significant improvements were achieved, making the robots more adaptable to rough terrains and representing a step further towards future applications and deployments.
Article
Computer Science, Information Systems
Xuan Shao, Ying Shen, Lin Zhang, Shengjie Zhao, Dandan Zhu, Yicong Zhou
Summary: This article introduces a large benchmark dataset called BeVIS for evaluating the performance of SLAM systems for autonomous indoor parking. The dataset includes both raw data and groundtruth trajectories collected from visual, inertial, and surround-view sensors. Additionally, a semantic SLAM framework called VISSLAM-2 is proposed for modeling various semantic objects on the ground. The effectiveness of VISSLAM-2 is demonstrated through experiments on the BeVIS dataset.
ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Yang Chen, Lin Zhang, Ying Shen, Brian Nlong Zhao, Yicong Zhou
Summary: An SVS consists of four fisheye cameras mounted around the vehicle for sensing the surrounding environment. A top-down surround-view can be synthesized from synchronized camera images, assuming the calibration of intrinsics and extrinsics. We propose a novel extrinsic self-calibration scheme, WESNet, which follows a weakly supervised framework to fill the research gap in extrinsic calibration.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Engineering, Electrical & Electronic
Sandip Bhattacharya, Mohammed Imran Hussain, John Ajayan, Shubham Tayal, Louis Maria Irudaya Leo Joseph, Sreedhar Kollem, Usha Desai, Syed Musthak Ahmed, Ravichander Janapati
Summary: In this study, a 6T-SRAM cell was designed using a 16-nm CMOS process. The performance in terms of read-speed latency was analyzed, and the use of temperature-dependent Cu and MLGNR-based nanointerconnect materials was investigated. The results showed that the read speed latency increased rapidly with the increase in interconnect length, with Cu interconnects having higher latency compared to GNR interconnects.
Article
Engineering, Electrical & Electronic
Tingting Xu, Xiaoyu Kong, Qiangqiang Shen, Yongyong Chen, Yicong Zhou
Summary: We propose a unified model under the plug-and-play framework that integrates deep prior and low-rank quaternion prior (DLRQP) for color image processing, addressing the limitations of traditional methods in representing color images holistically and flexibly fusing deep and handcrafted priors.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Automation & Control Systems
Hua Chen, Yu Jiang, Xiaogang Zhang, Yicong Zhou, Lianhong Wang, Jinchao Wei
Summary: In this article, a multivariate time series forecasting model based on dynamic spatio-temporal graph attention network (GAT) is proposed to model the time-varying spatio-temporal correlation between process data and perform long-range forecasting. The method includes an adaptive adjacency matrix generation algorithm to construct a basic graph structure for the process data and a spatio-temporal graph attention module for extracting spatial and temporal features. The results based on actual data demonstrate high prediction accuracy and wide application prospects in industrial processes.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Computer Science, Theory & Methods
Shen Wang, Zhaoyang Zhang, Guopu Zhu, Xinpeng Zhang, Yicong Zhou, Jiwu Huang
Summary: With the widespread use of automated speech recognition systems, attacks against these systems have become popular. Existing black-box attack methods for ASR systems are query-intensive and lack efficiency. This paper proposes a new black-box attack called MGSA, which generates adversarial audio samples with substantially fewer queries.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)
Article
Biochemical Research Methods
Shubham Tayal, Budhaditya Majumdar, Sandip Bhattacharya, Sayan Kanungo
Summary: The design of a high-performance DMFET with small device dimension has attracted significant research attention for POC diagenesis applications. A DM-JLNFET architecture is introduced and investigated for label-free electrochemical biosensing, showing superior sensing performance compared to conventional counterparts. The underlying physics of the transduction mechanism and the sensing performance are analyzed based on device simulations.
IEEE TRANSACTIONS ON NANOBIOSCIENCE
(2023)
Article
Engineering, Electrical & Electronic
Shuai Shao, Lei Xing, Yanjiang Wang, Baodi Liu, Weifeng Liu, Yicong Zhou
Summary: This study proposes a multi-view feature collaboration method to address the problem of data scarcity. It denoises multi-view features using a subspace learning method, and designs three attention blocks to balance the representation between different views. Experimental results on four benchmark datasets show significant improvements compared to previous methods.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Yulan Zhang, Guopu Zhu, Xing Wang, Xiangyang Luo, Yicong Zhou, Hongli Zhang, Ligang Wu
Summary: In this paper, a CNN-T GAN is proposed to distinguish between the source and target regions in copy-move forgery images. The generator generates a mask similar to the groundtruth mask, and the discriminator discriminates true and false image pairs. The proposed method improves the performance of copy-move forgery localization.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Min Shi, Jialin Shen, Qingming Yi, Jian Weng, Zunkai Huang, Aiwen Luo, Yicong Zhou
Summary: This article introduces a lightweight multiscale-feature-fusion network (LMFFNet) that achieves a good balance between accuracy and inference speed in real-time semantic segmentation. The network extracts features with fewer parameters, fuses multiscale semantic features to improve segmentation accuracy, and recovers details of input images through the attention mechanism. Experiments demonstrate that the proposed network is suitable for autonomous driving and robotics.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Engineering, Electrical & Electronic
Min Du, Lianhong Wang, Yicong Zhou
Summary: In this paper, a single mixed-integer linear programming model is presented for high-stealth false data attacks (FDAs) on overloading a set of lines by injecting stealthy false data. The proposed model reveals that an intelligent attacker is able to deliberately construct a valid attack vector to overload multiple transmission lines while hiding it among normal data to evade advanced anomaly detection methods. In addition, the proposed cyber-attack mode can help the attacker optimally select the targeted lines. Simulation results on multiple large-scale test systems validate the effectiveness of the proposed approach.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Engineering, Electrical & Electronic
Xiaolin Yin, Shaowu Wu, Ke Wang, Wei Lu, Yicong Zhou, Jiwu Huang
Summary: This paper proposes an anti-rounding image steganography method using a separable fine-tuning network architecture. The method addresses the challenge of rounding operation in steganographic applications by using joint training and fine-tuning stages. Experimental results demonstrate the robustness of the proposed method to rounding operation and its effectiveness in reducing degradation of image quality and steganalysis performance.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Zheng Zhou, Yongyong Chen, Yicong Zhou
Summary: Motivated by deep learning methods, this study proposes a deep unfolding method for image denoising tasks. The method inherits the advantages of both deep learning and traditional machine learning, and addresses the issues in existing methods. Experimental results demonstrate the superiority of the proposed method.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Zheng Zhou, Yue Wu, Yicong Zhou
Summary: This paper introduces a method called Consistent Arbitrary Style Transfer (CAST) that quantifies the consistency of generated images and effectively transfers style patterns while preserving consistency. The proposed CAST method incorporates IoUPC module, SA module, and SILR module to achieve consistent style transfer. Experimental results demonstrate that the CAST framework can effectively transfer style patterns while preserving consistency and achieves state-of-the-art performance.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Yongyong Chen, Xiaojia Zhao, Zheng Zhang, Youfa Liu, Jingyong Su, Yicong Zhou
Summary: Multiview clustering (MVC) has gained attention in recent years for its ability to uncover intrinsic clustering structures in data. However, previous methods only focus on either complete or incomplete multiview, lacking a unified framework to handle both tasks simultaneously. To address this, we propose a unified framework, called TDASC, which integrates tensor learning and dynamic anchor learning for scalable clustering. TDASC efficiently learns smaller view-specific graphs using anchor learning and incorporates multiple graphs into an inter-view low-rank tensor, effectively modeling high-order correlations across views. Experimental results on complete and incomplete multiview datasets demonstrate the effectiveness and efficiency of TDASC compared to state-of-the-art techniques.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)