Wuhan University · Expected Graduation
Research interests: Latent Action Modeling, Vision-Language-Action Models, Game AI, RAG Systems
Wuhan University · GPA: 3.89/4.0, Top 5%
Honors: Multiple First-Class Academic Scholarships
Xiaoyu Chen*, Hangxing Wei*, Pushi Zhang, et al.
Grounded latent actions in robot dynamics for VLA policy learning, improving simulation and real-world performance with zero-shot latent planning.
Hangxing Wei, Xiaoyu Chen, Chuheng Zhang, Tim Pearce, et al. · ICML 2026
Proposed an additively compositional latent space that filters non-essential environmental features for robust robot control.
Rushuai Yang*, Hangxing Wei*, Ran Zhang*, et al. · ECCV 2026
Proposed a diffusion-based RL method for collecting diverse, high-quality synthetic data in simulation environments.
Standardized access to 90+ datasets across 10+ robotic platforms and optimized cloud-based streaming for 50M+ demonstration episodes.
Designed a hardware abstraction stack and data collection pipeline for synchronized robot telemetry and observations.
Built config-centric training, evaluation, and alerting infrastructure for mainstream VLA models.
Led the vision and sensor-fusion system for a national champion robotics team, combining LiDAR, depth cameras, YOLO, and optical flow.
Architected an enterprise parsing framework for OCR, text extraction, table recognition, and reading-order reconstruction.
News · Code · Certificate
Multiple academic scholarships, Merit Student, and Outstanding Student awards.
Systems for AI · Research Engineering · Python · C/C++ · Linux · Docker · FastAPI · Azure Blob · vLLM · Distributed Training · PyTorch · ROS · OpenCV · Reinforcement Learning · Diffusion · VLA Models