Yichen Wang

I am a Research Assistant at the PAIRS Lab, The Hong Kong University of Science and Technology (Guangzhou), advised by Prof. Fangqiang Ding. I received my B.Eng. in Automation from Tongji University in June 2026.

My research lies at the intersection of Embodied AI and robotics. I am particularly interested in building learning-based policies that connect perception and action in complex, partially observable environments.

• Embodied AI — robot navigation, embodied question answering, Vision-Language-Action models, and world models.
• Spatial Perception — visual SLAM, multi-sensor fusion, and long-term spatial memory.
• Machine Learning — deep reinforcement learning, self-supervised learning, and graph neural networks.

For my undergraduate thesis, advised by Prof. Chao Huang, I developed a context-aware reinforcement learning method for mapless navigation. By combining graph-based spatial encoding with temporal memory and intrinsic curiosity, the full model achieved an average success rate of 0.84 and SPL of 0.66 across four simulated environments.

Previously, advised by Dr. Jia Xu, I led perception and planning work for an autonomous hexapod robot and developed a Transformer-enhanced reinforcement learning framework for electric bus fleet dispatch. See the research page for details.