A5056154880- Soft Robotics and Applications
- Image Enhancement Techniques
- Advanced Vision and Imaging
- Advanced Sensor and Energy Harvesting Materials
- Micro and Nano Robotics
- Robot Manipulation and Learning
- Underwater Vehicles and Communication Systems
- Image and Signal Denoising Methods
- Advanced Materials and Mechanics
- Mechanical Engineering and Vibrations Research
- Modular Robots and Swarm Intelligence
- Video Surveillance and Tracking Methods
- Obstructive Sleep Apnea Research
- Machine Learning and Algorithms
- DNA and Biological Computing
- Advanced Algorithms and Applications
- Robotics and Sensor-Based Localization
- Effects of Vibration on Health
- Dielectric materials and actuators
- Shape Memory Alloy Transformations
- Robotic Locomotion and Control
- Advanced Image Processing Techniques
- Soil Mechanics and Vehicle Dynamics
- Grey System Theory Applications
- Optical measurement and interference techniques
Shanghai Jiao Tong University
2010-2022
Beihang University
2008-2021
Beijing Advanced Sciences and Innovation Center
2018
Beijing Institute of Technology
2016
Collecting in shallow water (water depth: ~30 m) is an emerging field that requires robotics for replacing human divers. Soft robots have several promising features (e.g., safe interaction with the environments, lightweight, etc.) performing such tasks. In this article, we developed underwater robotic system a three-degree-of-freedom (3-DoF) soft manipulator spatial delicate grasping water. First, present design and fabrication of opposite-bending-and-stretching structure (OBSS). Then,...
We described a challenge named "Diabetic Retinopathy (DR)-Grading and Image Quality Estimation Challenge" in conjunction with ISBI 2020 to hold three sub-challenges develop deep learning models for DR image assessment grading. The scientific community responded positively the challenge, 34 submissions from 574 registrations. In we provided DeepDRiD dataset containing 2,000 regular images (500 patients) 256 ultra-widefield (128 patients), both having quality grading annotations. discussed...
Low visual quality has prevented underwater robotic vision from a wide range of applications. Although several algorithms have been developed, real time and adaptive methods are deficient for real-world tasks. In this paper, we address difficulty based on generative adversarial networks (GAN), propose GAN-based restoration scheme (GAN-RS). particular, develop multibranch discriminator including an branch critic the purpose simultaneously preserving image content removing noise. addition to...
This paper presents a soft actuator embedded with two types of eutectic alloys which enable sensing, tunable mechanical degrees freedom (DOF), and variable stiffness properties. To modulate the actuator, we low melting point alloy (LMPA) in bottom portion actuator. Different sections LMPA could be selectively melted by Ni–Cr wires twined underneath. acquire curvature information, EGaIn (eutectic gallium indium) was infused into microchannel surrounding chambers Systematic experiments were...
Abstract Soft grippers based on fluidic elastomer actuators have the characteristics of gentle and adaptable grasping that is difficult to realize by rigid grippers. However, it remains challenging implement a compact gripping device has multiple bending configurations exert appropriate force, sensory capabilities evaluate state. Here, we present soft gripper with variable effective lengths (VELs) achieved rapidly softening selective shape memory polymer sections (within 0.6 s) via flexible...
Collecting seafood animals (such as sea cucumbers, echini, scallops, etc.) cultivated in shallow water (water depth: ~30m) is a profitable and an emerging field that requires robotics for replacing human divers. Soft have several promising features (e.g., safe contact with the objects, lightweight, performing such task. In this paper, we implemented soft manipulator opposite-bending-and-extension structure. A simple rapid inverse kinematics method proposed to control spatial location...
In this paper, a nondominated sorting genetic algorithm (NSGA)-II-based calibration for the underwater binocular vision measurement system is proposed. The results will be incorrect with camera model in air. To deal problem, refractive and an akin triangulation are proposed to establish nonlinear relationship housing parameters between object its corresponding image plain points. A novel usage of checkerboard based on relative position corners employed set three optimal goals, i.e., distance...
In this paper, we present the design, fabrication and function of a soft actuator embedded with Low-Melting-Point Alloy (LMPA). By melting metal via joule heating, phase transformed from solid state to liquid state, by which stiffness changed over nearly an order magnitude. Through reheating-recrystallizing circle, can self-healing recovered crack state. The speed under different electrical currents was measured. Besides, two experiments were conducted investigate property enhancement...
Abstract In order to improve the working ability of continuum robots in confined space, a novel central support structure with spiral cutting pattern is proposed this paper. The hollow scheme was selected through analysing different patterns thin-walled superelastic Ni-Ti shape memory alloy (SMA) tube. Then laser process and heat treatment for fabricating were analysed. analysis results their impact on material characteristics proves that ensure performance material. single-segment...
This paper presents a soft actuator embedded with conductive liquid metal and shape memory epoxy (SME) which function together to enable self-sensing, tunable mechanical degrees of freedom (DoF), variable stiffness. We thermoplastic in the bottom portion actuator. Different sections SME could be selectively softened by an implanted silver yarn located at different positions. When electric current passes through yarn, it induces phase transition that changes from stiff state compliant state....
In this paper, a real-time underwater robotic visual tracking strategy (RUTS) based on image restoration and Kernelized Correlation Filters (KCF) is developed for robots. A unsupervised advancement scheme (RUAS), which utilized in strategy, performs robustly restoring images. The KCF, as high-speed accurate method land, employed strategy. To handle the conflict between speed accuracy, we propose KCF video sequence restored by RUAS, comparing Histogram of Oriented Gradient (HOG) descriptors...
The posture of a humanoid robot needs to be changed from lying state others in order complete required missions. With the rolling motion, it is possible for change and adapt itself environmental uncertainties. study motion robots fairly rare. This paper proposed highly dynamic whole-body motion. planned trajectories were based on human rolling, meet constraints stability multi-points contact conditions. trajectory center mass (CoM) calculated by dividing into upper lower body movement....
iQIYI-VID-2019 is the largest video dataset for multi-modal person identification. It composed of more than 200k clips 10,034 celebrities. Face a critical clue identification when face visible in video. However quality may not always be good, and it also contains lot noise caused by detection feature extraction. Meanwhile, conventional classification methods do fully exploit ability modality. They make full use confidence evaluation indicators, which are key information To address these...
This paper proposes a simple yet effective strategy for improving deep supervised learning, named Stochastic Label Refinery (SLR), by refining training labels to more informative labels. When neural network, target distributions (or ground-truth) are typically "hard", which means the label of each category consists only 0 and 1. However, fixed "hard" do not capture association between categories or that objects. In this study, instead using hard distributions, we iteratively generate "soft"...
Low visual quality has prevented underwater robotic vision from a wide range of applications. Although several algorithms have been developed, real-time and adaptive methods are deficient for real-world tasks. In this paper, we address difficulty based on generative adversarial networks (GAN), propose GAN-based restoration scheme (GAN-RS). particular, develop multi-branch discriminator including an branch critic the purpose simultaneously preserving image content removing noise. addition to...
The output prediction of photovoltaic (PV) power station is necessary because the weather, environment and seasonal factors lead to unstable PV generation, which will affect planning scheduling system. Compared with single model method, linear combination method could improve accuracy station. However, forecast a simple convex different methods lack general applicability. This paper presents nonlinear based on BP neural network ARMA predict plant. relationship between results two models...
Cryogenic grinding has been the development in field of ultrafine plastics, especially ones with low softening temperature. The common refrigeration methods are difficult to apply engineering production, as liquid nitrogen effects high operation costs. Freon, lithium bromide, and ammonia show dissatisfactory effect need complex system. authors chose a kind typical thermoplastic elastomer, EVA (ethylene vinyl acetate), their subject for research. Based on thorough study material properties...