Collective experience learning improves robot policies from fleet failures
Original title: Learning from Fleet Experience: A Collective Learning Paradigm for Robots
Held-out robots reach 93.3–99.2% success on five tasks; a 96-robot phone-packaging run rises from 79.2% to 98.3%.
Collective experience learning lets deployed robots share execution traces so a common policy improves without new expert demonstrations or teleoperator corrections. Failure-guided experience retrieval fine-tuning pulls relevant successful trajectory segments and uses their actions as supervision. A fault-tolerant asynchronous system coordinates collection, optimization, and policy distribution despite local interruptions. With per-robot collection time fixed, larger fleets generally yield higher success.
- FERT uses failure contexts to retrieve matching successful trajectory segments.
- Those actions supervise updates without additional corrective demonstrations.
- Each of five manipulation tasks was evaluated over 120 trials on excluded robots.
- Larger fleets generally raise success when collection time per robot is fixed.
From the original article
1 min read at huggingface.co
Experience generated during deployment offers an opportunity for robots to improve policy reliability through learning from their own executions. However, learning through human corrections requires teleoperators to monitor execution and intervene when policies fail, a demand that is difficult to meet as robot fleets grow.
By Chao Yu, Shu'ang Yu, Hao Lin · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.