A5101729358- Analytical chemistry methods development
- Laser-induced spectroscopy and plasma
- Hydraulic Fracturing and Reservoir Analysis
- Enhanced Oil Recovery Techniques
- Robot Manipulation and Learning
- Cultural Heritage Materials Analysis
- Hydrocarbon exploration and reservoir analysis
- Soft Robotics and Applications
- Drilling and Well Engineering
- Advanced Data Storage Technologies
- Hand Gesture Recognition Systems
- Geotechnical Engineering and Soil Stabilization
- Fault Detection and Control Systems
- Advanced Neural Network Applications
- Mercury impact and mitigation studies
- Carbon Dioxide Capture Technologies
- Earthquake Detection and Analysis
- Innovative concrete reinforcement materials
- Machine Fault Diagnosis Techniques
- Adversarial Robustness in Machine Learning
- Icing and De-icing Technologies
- Wind Energy Research and Development
- NMR spectroscopy and applications
- Manufacturing Process and Optimization
- Poverty, Education, and Child Welfare
China Railway Design Corporation (China)
2025
Yanshan University
2020-2024
Chinese Academy of Sciences
2016-2024
University of Chinese Academy of Sciences
2016-2024
Beijing Academy of Artificial Intelligence
2022-2024
Shandong Institute of Automation
2022-2024
Second Affiliated Hospital of Zhejiang University
2024
National University of Defense Technology
2023
South China University of Technology
2023
Institute of Automation
2023
With the rapid development of construction industry, waste is also increasing day by day. Considering utilization rate recycled coarse aggregate (RCA) concrete and mechanical strength RCA itself, this study carried out collaborative modification through new modifiers carboxylic butadiene-styrene latex (CSBL) PVA fiber, aiming at improving properties RCA. In study, compressive test shear were used to block. Scanning electron microscope (SEM) Fourier transform infrared spectrometer (FTIR)...
In this letter, we propose a novel Grasp Pose Domain Adaptation Network (GPDAN) to achieve sim-to-real domain adaptation for 6-DoF grasp pose detection. The main task of GPDAN is detect feasible poses in cluttered scenes. A point-wise self-supervised classification module with point cloud mixture and feature fusion strategy proposed as the auxiliary promote alignment between source target through adversarial training. Experimental results on both simulation real-world environments...
In unstructured environments, robotic grasping tasks are frequently required to interactively search for and retrieve specific objects from a cluttered workspace under the condition that only partial information about target is available, like images, text descriptions, 3D models, etc. It great challenge correctly recognize targets with limited learn synergies between different action primitives grasp densely occluding efficiently. this paper, we propose novel human-like push-grasping method...
In the microanalysis of laser-induced breakdown spectroscopy, influence surface roughness on spectral stability and quantitative analysis capability was studied for first time when laser ablation crater diameter approximately 10 μm.
Taking the spectrum generated by pure sample as matrix background and subtracting intensity value of at analysis line can reduce interference effects.
This paper presents a method for the automatic adjustment of laser defocusing amount in micro-laser-induced breakdown spectroscopy. A microscopic optical imaging system consisting CCD camera and 20× objective lens was adopted to realize method. The real-time auto-focusing achieved by detecting effective pixels light spot generated pointer. focusing accuracy could achieve 3 μm. element concentrations Mn Ni low-alloy steels were analyzed at crater diameter about 35 μm using presented After...
MapReduce, as a programming model, has become an important solution to large-scale data-intensive processing. It been widely used in various fields such Web search, machine learning and e-commerce. Hadoop, open-source implementation of is for offline massive data job. consists MapReduce HDFS. In the study we found parallel Hadoop coarse grained, it cannot take full advantage multi-core system. Eventually, this would lower utilization efficiency whole cluster. To improve into fine grained...
In this paper, we propose a novel Voting based Grasp Pose Network (VGPN) to detect 6-DoF grasps in cluttered scenes. The motivation of paper is that local object geometry can provide useful clues about where the be grasped. Generated by sampled seed points from raw point cloud, votes allow different regions contribute locations Geometric features various are aggregated generate more confident and dense space, which enables grasp prediction utilizing global context features. search space pose...
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
Utilizing Vision-Language Models (VLMs) for robotic manipulation represents a novel paradigm, aiming to enhance the model's ability generalize new objects and instructions. However, due variations in camera specifications mounting positions, existing methods exhibit significant performance disparities across different platforms. To address this challenge, we propose RoboUniView paper, an innovative approach that decouples visual feature extraction from action learning. We first learn unified...
Foundation models hold significant potential for enabling robots to perform long-horizon general manipulation tasks. However, the simplicity of tasks and uniformity environments in existing benchmarks restrict their effective deployment complex scenarios. To address this limitation, paper introduces \textit{RoboCAS} benchmark, first benchmark specifically designed object arrangement scenarios robotic manipulation. This employs flexible concise scripted policies efficiently collect a diverse...
In recent years, robotics has advanced significantly through the integration of larger models and large-scale datasets. However, challenges remain in applying these to 3D spatial interactions managing data collection costs. To address issues, we propose multimodal robotic manipulation model, RoboMM, along with comprehensive dataset, RoboData. RoboMM enhances perception camera parameters occupancy supervision. Building on OpenFlamingo, it incorporates Modality-Isolation-Mask decoder blocks,...
The seismic production technique is a physical method used to stimulate flow velocity nearby wellbore during oil field development. A numerical study on the working principle of this helps assess extent fluid variation. Previous works were mainly done with one-dimensional or radial models and ground vibration source hypothezied load earth. three-dimensional model as well downhole was better explore variation seepage considering real device. built quarter five-spot pattern. influences...