China's first robot doctoral student 'Xueba 01' enrolls at Shanghai Theatre Academy

【CMHnews】A robot doctoral candidate named "Xueba 01" recently enrolled at the Shanghai Theatre Academy (STA), media reported on Tuesday.

    The robot stands at an impressive 1.75 meters tall, weighing approximately 30 kilograms. It is China's first robot doctoral student. Xueba in Chinese means a top student or an academic star.

    Robot doctoral student Xueba 01 (C) stands between two students at the Shanghai Theatre Academy. (Photo/Shanghai Observer)

    Featuring an elegant appearance with fluid movements capable of physical interaction, its body is an optimized version of the "Xingzhe II (DroidUp X02-Lite)" robot, which won third place in the world's first humanoid robot half-marathon held in April in Beijing, developed by Professor Li Qingdu's team at the Shanghai University of Technology.

    With similar facial expressions to humans, and skin made of silicone that feels like elastic, it bears a huge resemblance to humans.

    In the classroom, it interacted with CCTV News host Pan Tao, who is also a newly enrolled 2025 doctoral candidate.

    "Its reactions are exceptionally swift. Interacting with it involved both verbal and eye contact, which was totally beyond my imagination," Pan said.

    After its studies, it joined classmates to exercise at the gym.

    Its supervisor Yang Qingqing will lead efforts to develop its "core performance," ensuring it possesses performance abilities and creative works.

    Li said "Xueba 01" can currently perform Yu Opera (Henan opera), Hu Opera (Shanghai Opera), and Peking Opera (Beijing opera), though its performance still needs to be fine-tuned.

    However, its movements do lack coherence and aesthetic appeal, which will require further training by learning from some professional performers.

    Initially, the developers built datasets using pure "motion capture" methods, whereby they invited Peking Opera actors to perform, capturing their movements, expressions, and gestures, which were used to then train the robot.

    In the future, it may learn directly by "watching videos" by itself as video datasets offer larger corpora, potentially enabling more natural performances, Li said.

Editor: Mo Honge

Reposting from ECNS.Cn