About Me
I am a robotics researcher working toward cognitive autonomy—robots that not only act, but understand, adapt, and self-correct. I am currently building the first step with TIWM (Tokenized Intent World Model), a sparse intent-token interface between perception and decision-making. My long-term goal is to unify interpretable cognitive templates and autoregressive long-horizon planning into a hybrid paradigm for generalizable, human-aligned robotic intelligence.
- Cognitive Robotics
- World Representation
- Deep Reinforcement Learning
- Robotic Middleware (ROS/ROS2)
- LLM/VLM for Robotics
MEng Computer Technology
Central China Normal University
MSc Computer Science
University of Wollongong
BSc Computer Science and Technology
Huaiyin Institute of Technology
- Sparse World Representation
Learning minimal yet semantically grounded tokens from high-dimensional perception to replace dense world-model reconstruction and enable fuzzy-memory sparse alignment across scenes.
- Cognitive Skill Templates
Designing interpretable, composable action schemas (symbolic abstraction × RL policy) for contextual reasoning, fast adaptation, and robust error recovery in dynamic human–robot interaction.
- Intent Autoregression & Spatiotemporal Co-evolution
Imagining future states via intent-token autoregression to support long-horizon, joint spatiotemporal planning without dense future rollouts.
