A5002625448- Manufacturing Process and Optimization
- 3D Shape Modeling and Analysis
- Robotics and Sensor-Based Localization
- Robotic Path Planning Algorithms
- Autonomous Vehicle Technology and Safety
- Computational Geometry and Mesh Generation
- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Computer Graphics and Visualization Techniques
- Digital Image Processing Techniques
- Evacuation and Crowd Dynamics
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Medical Imaging and Analysis
- Relativity and Gravitational Theory
- UAV Applications and Optimization
- Reinforcement Learning in Robotics
- Metal Forming Simulation Techniques
- Industrial Vision Systems and Defect Detection
- Myeloproliferative Neoplasms: Diagnosis and Treatment
- Engineering Technology and Methodologies
- Radiomics and Machine Learning in Medical Imaging
- Cosmology and Gravitation Theories
- Image and Object Detection Techniques
- Maritime Navigation and Safety
Soochow University
2024
First Affiliated Hospital of Soochow University
2024
Central South University
2021-2024
Liaoning Technical University
2024
China Agricultural University
2024
Ministry of Agriculture and Rural Affairs
2024
PolyGene
2024
Harbin Institute of Technology
2023
Southwest Petroleum University
2021
The Ohio State University
1995-1996
Neuromorphic computing has received more and attention recently since it can process information interact with the world like human brain. Agriculture is a complex system that includes many processes of planting, breeding, harvesting, processing, storage, logistics, consumption. Smart devices in association artificial intelligence (AI) robots Internet Things (IoT) systems have been used also need to be improved accommodate growth computing. great potential promote development smart...
Autonomous exploration in unknown environments is a fundamental task of Unmanned Aerial Vehicles (UAVs). To choose goals wisely, we propose an information-driven strategy by applying the fast marching method to UAVs. A frontier point detection algorithm designed obtain Candidate Goals (CGs) utilizing structural characteristics octree-based map. With sum information gain during journey as evaluation indicator, present novel utility function evaluate CGs considering trade-off between and...
In this work, we seek to predict camera poses across scenes with a multi-task learning manner, where view the localization of each scene as new task. We propose OFVL-MS, unified framework that dispenses traditional practice training model for individual and relieves gradient conflict induced by optimizing multiple collectively, enabling efficient storage yet precise visual all scenes. Technically, in forward pass design layer-adaptive sharing policy learnable score layer automatically...
Abstract Autonomous exploration is a critical technology to realize robotic intelligence as it allows unsupervised preparation for future tasks and facilitates flexible deployment. In this paper, novel Deep Reinforcement Learning (DRL) based autonomous strategy proposed efficiently reduce the unknown area of workspace provide accurate 2D map construction mobile robots. Different from existing human-designed techniques that usually make strong assumptions about scenarios tasks, we utilize...
The Object goal Navigation (ObjectNav) task requires an agent to navigate through a previously unknown domestic scenario using spatial and semantic contextual information, where the is specified by label (e.g., find TV). Such especially challenging as it formulating understanding complex co-occurrence relations among objects in diverse settings, which critical for long-sequence navigational decision-making. Existing methods learn either explicitly represent relationships discrete priors, or...
We present CO-Net, a cohesive framework that optimizes multiple point cloud tasks collectively across heterogeneous dataset domains. CO-Net maintains the characteristics of high storage efficiency since models with preponderance shared parameters can be assembled into single model. Specifically, we leverage residual MLP (Res-MLP) block for effective feature extraction and scale it gracefully along depth width network to meet demands different tasks. Based on block, propose novel nested...
Because of the nature die casting process, part geometry severely restricts and hence affects quality part. However, as is often case in other manufacturing processes, diecastings are currently designed purely based on their function. The manufacturability not considered until design has been nearly completed detailed. This due to support limitations current CAE tools. In this paper, we present a new volume-based approach diecastability evaluation, especially preliminary design. Our can be...
The tedious process of building an input model and manual communication between CAD Finite Element Analysis (FEA) significantly restricts the designers from using FEA, particularly in preliminary design. Most previous works concentrate on expediting this by interfacing FEA. This paper propose full integration FEA simulation code within a system, thus several problems are avoided. section-based is used as better suited to early It requires, comparing 3-D development additional functions...
Binary black hole mergers affect many different things inside the galaxy. Only recently, with LIGO developments, have we observed binary from observing gravitational wave events. Both common envelope theory and chemically homogenous are theories that explain how holes merge. It is unclear as to which true explanation for mergers. Throughout paper, two models, they work, their flaws, evidence in support of them explored. The paper aims answer question: What there 2 types merger models?
The development of Artificial Intelligence (AI) Large Models has a great impact on the application automotive Intelligent cockpit. fusion Cockpit and become new growth point user experience in industry, which also creates problems for related scholars, practitioners users terms their understanding evaluation capability characteristics (ICLM). This paper aims to analyse current situation cockpit, large model AI Agent, reveal key research focuses integration Models, put forward necessary...
End-to-end autonomous driving offers a streamlined alternative to the traditional modular pipeline, integrating perception, prediction, and planning within single framework. While Deep Reinforcement Learning (DRL) has recently gained traction in this domain, existing approaches often overlook critical connection between feature extraction of DRL perception. In paper, we bridge gap by mapping network directly perception phase, enabling clearer interpretation through semantic segmentation. By...
Abstract Because of the nature die casting process, part geometry severely restricts and hence affects quality part. However, as is often case in other manufacturing processes, diecastings are currently designed purely based on their function. The manufacturability not considered until design has been nearly completed detailed. This due to support limitations current CAE tools. In this paper, we present a new volume-based approach diecastability evaluation, especially preliminary design. Our...
In many mechanical design-related activities, the visualization tool needs to convey not only shape of objects, but also their interior problem regions. Due binary nature these models, existing shading models often fall short supporting a realistic display. this case study, we present several new contextual methods that originally developed for our design tools. The results are then compared with gray-scale applied gray-level version object. comparison shows method can be any object and...
We present a new visualization approach to support design for manufacturing (DFM). This involves the correlation of problems with causative geometric characteristics. then discuss use distance transform and 3-D thinning extract these characteristics from voxelized object. In contrast current computer aided engineering (CAE) tools, our system is very efficient simple use. It does not require skill experience generate control numerical mesh interpret results. The specifically tailored makes...
We present a new visualization approach to support design for manufacturing (DFM). This involves the correlation of problems with causative geometric characteristics. then discuss use distance transform and 3-D thinning extract these characteristics from voxelized object. In contrast current computer aided engineering (CAE) tools, our system is very efficient simple use. It does not require skill experience generate control numerical mesh interpret results. The specifically tailored makes...
This study presents an identification and localization algorithm for transparent columnar lenses. For a given color photo of lens array, based on the principle Radon transform knowledge image processing, we construct generalized function that focuses recognition circular targets, finally achieve purpose recognizing obtaining coordinates radius circle center each lens. According to different functions, design this is divided two parts: first part preprocessing, second implementation user...
In this work, we seek to predict camera poses across scenes with a multi-task learning manner, where view the localization of each scene as new task. We propose OFVL-MS, unified framework that dispenses traditional practice training model for individual and relieves gradient conflict induced by optimizing multiple collectively, enabling efficient storage yet precise visual all scenes. Technically, in forward pass design layer-adaptive sharing policy learnable score layer automatically...