A5103064820- Robot Manipulation and Learning
- Reinforcement Learning in Robotics
- Topic Modeling
- Evaluation Methods in Various Fields
- Safety and Risk Management
- Web Data Mining and Analysis
- Soft Robotics and Applications
- Geographic Information Systems Studies
- Fire Detection and Safety Systems
- Neural Networks and Applications
- Natural Language Processing Techniques
- Traffic and Road Safety
- Advanced Graph Neural Networks
- Hand Gesture Recognition Systems
- Human Pose and Action Recognition
- Evacuation and Crowd Dynamics
- Modular Robots and Swarm Intelligence
- Intelligent Tutoring Systems and Adaptive Learning
Minzu University of China
2022-2024
University of Illinois Urbana-Champaign
2024
Northwest Minzu University
2023
Ministry of Education of the People's Republic of China
2022
Large, high-capacity models trained on diverse datasets have shown remarkable successes efficiently tackling downstream applications. In domains from NLP to Computer Vision, this has led a consolidation of pretrained models, with general backbones serving as starting point for many Can such happen in robotics? Conventionally, robotic learning methods train separate model every application, robot, and even environment. we instead generalist X-robot policy that can be adapted new robots,...
We present a package of annotation resources, including guideline, flowchart, and an Intelligent Tutoring System for training human annotators. These resources can be used to apply Rhetorical Structure Theory (RST) essays written by students in K-12 schools. Furthermore, we highlight the great potential using RST provide automated feedback improving writing quality across genres.
Based on the theoretical framework of Pearl's causal model, this study conducted research in three main areas to address lack work model construction urban fire domain. First, at association level, PC (Peter-Clark) algorithm was used construct an initial network domain establish relationships between relevant factors. However, due potential missing directed casual nodes, structure supplemented based knowledge provided by experts. Second, intervention applied backdoor criterion and front-door...
Videos of robots interacting with objects encode rich information about the objects' dynamics. However, existing video prediction approaches typically do not explicitly account for 3D from videos, such as robot actions and states, limiting their use in real-world robotic applications. In this work, we introduce a framework to learn object dynamics directly multi-view RGB videos by considering robot's action trajectories effects on scene We utilize Gaussian representation Splatting (3DGS)...
Scalable learning of humanoid robots is crucial for their deployment in real-world applications. While traditional approaches primarily rely on reinforcement or teleoperation to achieve whole-body control, they are often limited by the diversity simulated environments and high costs demonstration collection. In contrast, human videos ubiquitous present an untapped source semantic motion information that could significantly enhance generalization capabilities robots. This paper introduces...
In firefighting and rescue, the influence of various factors fire is very critical. To explore relationship between severity urban fires attributes fires, basic information a large number cases was collected sorted out, case data were defined explained strictly scientifically. The C4.5 CART algorithms in decision tree model used to mine analyze data, levels based on entropy, gain ratio, GINI coefficient established. prediction results verify correctness model. plays certain auxiliary role...
In recent years, urban fires have become a frequent topic of hot search, the casualties and economic losses caused by many discussions, are getting more attention. Forecasting number computer is beneficial to reduce provide assistance for arranging deployment fire police making decisions as soon possible. Based on prognostication accidents using gray model, Markov model introduced rectify residual error in prediction thereby enhancing its predictive accuracy. this paper, accident data...
Classical decision theory refers to the use of subjective probability estimation for some unknown states under incomplete information, occurrence is revised based on Bayesian formula, and then maximum expected value calculated according utility distribution plan. However, in field urban firefighting, few researchers have realized evaluation firefighting plans through classical theory. Aiming at this problem, paper constructs a decision-making network analysis fire protection data our...