Human-Robot Interaction
Using gaze, language, and interaction context to make robot policies easier for humans to steer in real tasks.
MARS Lab, Nanyang Technological University, Singapore
Kuangji Zuo 左匡济
Graduate researcher at MARS Lab, Nanyang Technological University.
My research focuses on embodied intelligence for humanoid robots, with an emphasis on vision-language-action learning, human intent understanding, and natural human-robot interaction.
Research
Embodied AI for natural human-robot interaction
Whole-Body VLA
Building vision-language-action systems that connect high-level instructions with physically grounded whole-body humanoid behavior.
Intent-Aware Autonomy
Studying humanoid systems that infer what people mean, not only what they explicitly say, during collaborative manipulation.
Publications
Selected publications and preprints
About
My research focuses on embodied intelligence for humanoid robots, especially whole-body VLA models and human-intention-aware robot learning. I am interested in systems where humans can guide robots through natural cues such as gaze, language, and interaction context.
I work on robot policies that connect high-level human intent with physically grounded action, with a focus on ambiguous interaction settings where gaze, language, and context can help robots choose the right behavior.
Search and research keywords associated with my work include embodied AI, humanoid robotics, human-robot interaction, VLA, whole-body VLA, whole-body action models, WAM, gaze-conditioned robot learning, and intent-aware robot policies.
Contact
For research conversations or collaboration, reach me at kuangji001@e.ntu.edu.sg . You can also find my academic and project updates through the links below.