Tsinghua University, together with Wuwen Xinqiong and Zhiyuan, proposed a
Collective Experience Learning (CEL) paradigm and demonstrated it on a 96-robot
physical swarm. CEL lets robots learn not only from human demonstrations but by
sharing and accumulating other robots’ successful experiences, enabling
continual, collective improvement. The team frames CEL as a technical path
toward autonomous evolution and scaled deployment of robotic systems.