Tsinghua, Infinigence AI and AgiBot propose Collective Experience Learning for 96 robots
Tsinghua University, Infinigence AI (无问芯穹) and AgiBot jointly released the Collective Experience Learning (CEL) paradigm, letting robots learn from peers' real task trajectories instead of relying only on human demonstrations. CEL pairs the FERT failure-guided experience-retrieval fine-tuning algorithm with the FATS fault-tolerant asynchronous training system, built on the open-source RLinf framework to coordinate up to 96 real robots. According to the project team, CEL surpassed the average performance of six human remote operators across five real tasks (flower arranging, folding clothes, phone charging, stool assembly and phone packaging), and on phone packaging the success rate rose from 86.7% with 16 training robots to 98.3% with 96.
The work targets a core bottleneck in embodied AI — the shortage of human demonstrators relative to deployed robot fleets — and demonstrates that per-robot experience can be pooled into a shared learning signal at the scale of a 96-robot cluster, though all figures come from the participating vendors and university team without independent replication.
从学习人类到学习机器人同伴，清华大学联合无问芯穹、智元提出具身智能规模化落地新路径 – 量子位
< img id="wx_img" src="https://www.qbitai.com/wp-content/uploads/imgs/qbitai-logo-1.png" width="400" height="400">
从学习人类到学习机器人同伴，清华大学联合无问芯穹、智元提出具身智能规模化落地新路径
量子位的朋友们 2026-10-30 11:48:04 来源：量子位
让机器人成为彼此的“老师”
#Collective Experience Learning (CEL)#Tsinghua University#Infinigence AI#AgiBot#FERT#FATS#RLinf