Tsinghua University, Wuwen Xinqiong, and Agibot Validate Collective Experience Learning Paradigm on 96 Humanoid Robots

Tsinghua University, along with AI infrastructure firm Wuwen Xinqiong and robotics company Agibot, have validated a new paradigm called Collective Experience Learning (CEL) using 96 real humanoid robots. This approach enables robots to share experiences from both successes and failures, reducing reliance on human teleoperation. In tests involving five tasks such as flower arranging and phone packaging, the robot group achieved an overall success rate of 97.5%, exceeding the average performance of six human operators. The system includes a Fault-Tolerant Asynchronous Training System (FATS) and a FERT algorithm to optimize strategies despite hardware faults.