Max Robotics — Global Service Robot Directory + US Certification

About

Junjia Liu is a pioneering robotics researcher whose work lies at the intersection of soft robotics, reinforcement learning, and dexterous manipulation. His research focuses on enabling robots to master complex, real-world tasks—from manipulating soft objects to playing the piano—by developing novel learning frameworks that overcome the limitations of traditional control methods. Liu’s most significant contributions include the introduction of SoftGPT, a generative pre-trained heterogeneous graph transformer that allows robots to learn goal-oriented soft object manipulation skills from human demonstrations, addressing the challenge of variable shape dynamics in domestic environments. His work on Mixline, a hybrid reinforcement learning framework, tackles long-horizon bimanual tasks like coffee stirring, while ReVoLT combines relational reasoning with Voronoi local graph planning for efficient target-driven navigation in unknown spaces. Liu has also advanced dexterous manipulation through his work on the Humanoid Pianist, which uses synergy-based hand representations to achieve fluent piano playing. With his papers accumulating citations and his innovative approaches to data-efficient learning, Liu is shaping the future of embodied AI and robotic skill acquisition.