Research Projects
Zhihao Xia's research focuses on AI for Science, including AI agents for scientific research, large language model post-training and domain-specific adaptation, and embodied intelligence for intelligent robotic systems. Zhihao Xia's earlier work spans mechanism synthesis, generalized parallel mechanisms, and humanoid robotics.
Research Portfolio
Humanoid Robotics
Participated in the development of Guanghua-0 and Guanghua-1, Fudan University's general-purpose humanoid robot prototypes. The work involved robotic mechanism integration, system development, and coordination between embodied-intelligence methods and physical robot design.
AI for Science
Develops AI agents for scientific reasoning and research workflows, with an emphasis on large language model post-training and domain-specific adaptation. Mechanism synthesis and intelligent robotics provide core scientific testbeds for this research.
Mechanism Design
Designed and analyzed a 6-DOF generalized parallel manipulator inspired by humanoid leg structures. The project connected topological synthesis, kinematic analysis, and humanoid robotic structure design.
Origami & Reconfiguration
Explores graph-theoretic and topology-based methods for synthesizing origami-inspired, continuous-origami, configurable, and reconfigurable mechanisms for robotic applications.
Aerial Manipulation
Investigates generalized parallel mechanisms for adaptive landing, aerial manipulation, buffer landing systems, and configurable robotic structures with improved motion and force transmission.
Early Research
Led a provincial undergraduate innovation project on a collaborative manipulator integrating mechanical design, embedded control, visual recognition, and human-machine interaction.