A5028749019- Target Tracking and Data Fusion in Sensor Networks
- Gaussian Processes and Bayesian Inference
- Advanced Control Systems Optimization
- Control Systems and Identification
- Distributed Sensor Networks and Detection Algorithms
- Stability and Control of Uncertain Systems
- Balance, Gait, and Falls Prevention
- Distributed Control Multi-Agent Systems
- Adaptive Control of Nonlinear Systems
- Robotics and Sensor-Based Localization
- Energy Efficient Wireless Sensor Networks
- Musculoskeletal pain and rehabilitation
- Aortic aneurysm repair treatments
- Fault Detection and Control Systems
- Robot Manipulation and Learning
- Iterative Learning Control Systems
- Modular Robots and Swarm Intelligence
- Structural Health Monitoring Techniques
- Indoor and Outdoor Localization Technologies
- Hydraulic and Pneumatic Systems
- Robotic Path Planning Algorithms
- Reinforcement Learning in Robotics
- Cardiac, Anesthesia and Surgical Outcomes
- Muscle activation and electromyography studies
- Advanced Vision and Imaging
Yonsei University
2001-2024
University of California, Berkeley
2001-2024
Kyungpook National University
2023
Leibniz University Hannover
2022
Michigan State University
2010-2021
Sungkyunkwan University
2014
Samsung Medical Center
2014
Lateral cephalometry has been widely used for skeletal classification in orthodontic diagnosis and treatment planning. However, this conventional system, requiring manual tracing of individual landmarks, contains possible errors inter- intravariability is highly time-consuming. This study aims to provide an accurate robust diagnostic system by incorporating a convolutional neural network (CNN) into 1-step, end-to-end with lateral cephalograms. A multimodal CNN model was constructed on the...
In this brief, we present a set of techniques for finding cost function to the time-invariant linear quadratic regulator (LQR) problem in both continuous- and discrete-time cases. Our methodology is based on solution inverse LQR problem, which can be stated as: does given controller K describe if so, what weights Q R produce as optimal solution? motivation investigating analysis motion goals biological systems. We first an efficient matrix inequality (LMI) method determining general case...
Abstract Background Despite the integral role of cephalometric analysis in orthodontics, there have been limitations regarding reliability, accuracy, etc. landmarks tracing. Attempts on developing automatic plotting systems continuously made but they are insufficient for clinical applications due to low reliability specific landmarks. In this study, we aimed develop a novel framework locating with confidence regions using Bayesian Convolutional Neural Networks (BCNN). Methods We trained our...
Tactile sensor arrays have attracted considerable attention for their use in diverse applications, such as advanced robotics and interactive human–machine interfaces. However, conventional tactile suffer from electrical crosstalk caused by current leakages between the cells. The approaches that been proposed thus far to overcome this issue require complex rectifier circuits or a serial fabrication process. This article reports flexible array fabricated through batch process using mesh. A...
In this paper, we consider mobile sensor networks that use spatiotemporal Gaussian processes to predict a wide range of physical phenomena. Nonparametric process regression is based on truncated observations proposed for with limited memory and computational power. We first provide theoretical foundation observations. particular, demonstrate prediction using all can be well approximated by under certain conditions. Inspired the analysis, then propose centralized navigation strategy move in...
This paper presents a novel class of self-organizing sensing agents that adaptively learn an anisotropic, spatio-temporal Gaussian process using noisy measurements and move in order to improve the quality estimated covariance function. approach is based on anisotropic functions processes introduced model broad range physical phenomena. The function assumed be unknown priori. Hence, it by maximum posteriori probability (MAP) estimator. prediction field interest then obtained MAP estimate An...
This study presents a new technology that can detect and discriminate individual chemical vapors to determine the vapor composition of mixed in situ based on multiplexed DNA-functionalized graphene (MDFG) nanoelectrode without need condense original or target dilution. To best our knowledge, artificial intelligence (AI)-operated arrayed electrodes were capable identifying compositions gases with ratio early stage. innovative comprised an optimized combination nanodeposited techniques...
An abdominal aortic aneurysm (AAA) is a gradual enlargement of the aorta that can cause life-threatening event when rupture occurs. Aneurysmal geometry has been proved to be critical factor in determining surgically treat AAAs, but, it challenging predict patient-specific evolution an AAA with biomechanical or statistical models. The recent success deep learning biomedical engineering shows promise for predictive medicine. However, model requires large dataset, which limits its application...
In this paper, we generated intelligent self-driving policies that minimize the injury severity in unexpected traffic signal violation scenarios at an intersection using deep reinforcement learning. We provided guidance on reward engineering terms of multiplicity objective function. used a deterministic policy gradient method simulated environment to train agents. designed two agents, one with single-objective function collision avoidance and other multi-objective both goal-approaching....
When designing a controller for the autonomous vehicle system, safety and trajectory tracking performance are two major concerns. This letter proposes novel control design an system with nonaffine inputs that can track desired trajectories while considering constraint. First, dynamics is modeled using differential flatness approach. The dynamic inversion method then employed of nonaffine-in barrier function (CBF) approach utilized to enforce handled as least-squares optimization problem, CBF...
In this paper, we formulate Gaussian process regression with observations under the localization uncertainty due to resource-constrained sensor networks. our formulation, effects of observations, measurement noise, uncertainty, and prior distributions are all correctly incorporated in posterior predictive statistics. The analytically intractable statistics proposed be approximated by two techniques, viz., Monte Carlo sampling Laplace's method. Such approximation techniques have been...
This brief presents a practical solution to the problem of monitoring an environmental process in large region by small number robotic sensors. Optimal sampling strategies are developed, taking into account quality estimated field and lifetime We also present experimental results for temperature outdoor swimming pool sampled autonomous aquatic surface robot. Simulation provided validate proposed scheme.
Inverse reinforcement learning (IRL) is a technique for automatic reward acquisition, however, it difficult to apply high-dimensional problems with unknown dynamics. This article proposes an efficient way solve the IRL problem based on sparse Gaussian process (GP) prediction l1-regularization only using highly limited number of expert demonstrations. A GP model proposed be trained predict function trajectory-reward pair data generated by deep different functions. The successfully predicts...
Background Osteoporosis is one of the diseases that requires early screening and detection for its management. Common clinical tools machine-learning (ML) models osteoporosis have been developed, but they show limitations such as low accuracy. Moreover, these methods are confined to limited risk factors lack individualized explanation. Objective The aim this study was develop an interpretable deep-learning (DL) model with features. Clinical interpretation individual explanations feature...
based on the teams design choices
In this technical note, we formulate a fully Bayesian approach for spatio-temporal Gaussian process regression such that multifactorial effects of observations, measurement noise and prior distributions are all correctly incorporated in the predictive distribution. Using discrete probabilities compactly supported kernels, provide way to design sequential prediction algorithms which exact can be computed constant time as number observations increases. For special case, distributed...