陳奕廷

Learning Human-Like Dexterous Hand Manipulation via Adversarial Motion Priors and Contact Guidance

資訊科學與工程研究所

多手指機器手、基於深度學習的機器人抓取與操作、強化學習、Multifingered Hands、Deep Learning in Grasping and Manipulation、Reinforcement Learning

A SLURM-based Orchestration System for Large-Scale Scenario-based Safety Validation of Automated Driving Systems

資訊科學與工程研究所

自駕車、情境式安全驗證、分散式系統、高效能運算、SLURM、容錯、autonomous vehicles、scenario-based safety validation、distributed systems、high-performance computing、fault tolerance

A Modular Closed-Loop Testing Framework for Scenario-Based Safety Validation of Automated Driving Systems: Planning and Control Evaluation

資訊科學與工程研究所

自動駕駛系統、情境式安全驗證、閉迴路測試、模組化框架、規劃與控制評估、Automated Driving Systems、Scenario-Based Validation、Closed-Loop Testing、Modular Framework、Planning and Control Evaluation

Learning to Imagine Normality for Enhancing Vision-Language Models in Industrial Anomaly Analysis

資訊科學與工程研究所

工業異常分析、視覺語言模型、正常性修復目標、細粒度視覺表徵、智慧型工業助理、Industrial Anomaly Analysis、Vision-Language Models、Normality Repair Objective、Fine-grained Visual Representations、Intelligent Industrial Assistant

Simulation-Based Spillage Predictor for Spillage-Aware Guided Diffusion Policies in Robot Food Scooping Tasks

多媒體工程研究所

引導式擴散策略、資料收集、機器人食物舀取模擬到真實轉移、溢出預 測器、模擬到真實轉移、溢出預測器、Guided Diffusion Policy、Data Collection、Robotic Food Scooping、Sim-to- Real Transfer、Spillage Predictor

Scenario-Based Autonomous Vehicle Safety Validation via Unified Transformer Encoder

數據科學與工程研究所

自動駕駛、驗證系統、模型泛化能力、場景語意關係、ODD、Autonomous vehicle、Safety validation system、Generalization ability

Controllable Collision Scenario Generation via Collision Pattern Prediction

機器人碩士學位學程・共同指導：邱維辰

自動駕駛、碰撞場景生成、資料集搜集、規劃器強化、安全驗證、Autonomous Driving、Safety Evaluation、Planner Improvement、Dataset Collection、Collision Scenario Generation

Weakly-Supervised Target Atomic Activity Localization in Traffic Scenes via Attention-Guided Pseudo Masks Refinement

數據科學與工程研究所

弱監督學習、動作定位、交通場景理解、電腦視覺、Weakly-supervised Learning、Action Localization、Traffic Scenes Under- standing、Computer Vision

Uncertainty-Aware Visual Risk Object Identification via Conformal Risk Tube Prediction

多媒體工程研究所

電腦視覺、智慧駕駛系統、共形預測、風險識別、深度學習、Computer Vision、Intelligent Driving System、Conformal Prediction、Risk Identification、Deep Learning

A Systematic Evaluation Framework for Modular Robotic Manipulation Policies via Structured Condition Space: A Case Study on Precision Pick-and-Place Tasks

數據科學與工程研究所

機器人操作、真實世界評估、系統化評估、可靠性分析、robot manipulation、real-world evaluation、systematic evaluation、reliability analysis

GUIDE: Generating Unobstructed Situated Subgoals via Predicting Goal-Aligned Centerlines for Trajectory Planning

多媒體工程研究所

場景化子目標、運動規劃、自動駕駛、目標對齊中心線、子目標選擇、Situated Subgoals、Motion Planning、Autonomous Driving、Goal-Aligned Centerlines、Subgoal Selection

Learning to Hang Everyday Objects with Deformed Semantic Keypoint Trajectory

網路工程研究所

機器人感知、電腦視覺、可操作之視覺表示、語意關鍵點軌跡、軌跡變形、機器人掛置任務、Robot Perception、Computer Vision、Actionable Visual Representation、Semantic Keypoint Trajectory、Trajectory Deformation、Hanging

Potential Fields as Scene Affordance for Behavior Change-Based Visual Risk Object Identification

多媒體工程研究所

智慧駕駛系統、場景可供性、勢場、風險物體辨識、因果推理、Intelligent Driving Systems、Scene Affordance、Potential Field、Risk Object Identification、Causal Inference

Task-Oriented Human Grasp Synthesis via Context- and Task-Aware Diffusers

資訊科學與工程研究所

人類抓取姿勢生成、擴散模型、手物體互動、場景理解、三維電腦視覺、Human Grasp Synthesis、Diffusion Models、Hand-Object Interaction、Scene Understanding、3D Computer Vision

SCONE: A Food Scooping Robot Learning Framework with Active Perception

數據科學與工程研究所

食物操作、機器人舀取任務、主動感知、演示學習、神經網路、Food Manipulation、Robot Scooping、Active Perception、Learn from Demonstration、Neural Networks