I’m an undergraduate at the Gaoling School of Artificial Intelligence, Renmin University of China.
I am passionate about Visual SLAM and Multimodal Reasoning, and I strive to develop the next generation of efficient, robust, and intelligent systems for real-world perception, localization, and mapping.
Connect with me:
My work focuses on building practical and high-performance systems for autonomous agents and Visual SLAM.
-
- A RAG conversational AI combining FAISS semantic retrieval with pluggable LLM backends (Claude, DeepSeek, or local Ollama). Features multi-turn session memory, SSE token streaming, a modern Vue 3 immersive Galgame UI, and similarity-threshold filtering for hallucination-free character reproduction.
-
- End-to-end lightweight C++ monocular visual odometry combining KLT optical flow tracking and robust pose estimation. Supports real-time camera trajectory visualization, demonstrating classical multi-view geometry for spatial perception applications.
-
Languages: C++17 · Python3 · Java · Go
-
SLAM & Perception: ORB-SLAM3 · Feature-based & Direct VO · VIO · Loop Closure · Dynamic Map Update
-
Optimization & Theory: Factor Graph · Bundle Adjustment · Lie Group/Algebra · EKF/ESKF
-
Deep Learning: SuperPoint · SuperGlue · Visual Transformers (ViT)
-
Libraries & Tools: Eigen · OpenCV · g2o · Ceres · ROS2 · CUDA