Learning from Fleet Experience: A Collective Learning Paradigm for Robots

Key points

• Experience generated during deployment offers an opportunity for robots to improve policy reliability through learning from their own executions.1
• We realize CEL through failure-guided experience retrieval fine-tuning (FERT), which uses failure contexts from robot executions to retrieve relevant trajectory segments from the fleet's shared successful experience and uses their actions as supervision for policy updates, without collecting additional expert demonstrations or corrective actions.1
• A fault-tolerant asynchronous training system (FATS) coordinates experience collection, policy optimization and policy distribution despite local interruptions within the robot fleet.1
• These results support collective experience as a practical basis for scalable robot learning and a path towards improving policy reliability during deployment without collecting additional expert corrective actions.1

Sources (2)

• [1]Learning from Fleet Experience: A Collective Learning Paradigm for Robots
  Hugging Face Daily Papers · Oct 19, 12:00 AM

  We introduce collective experience learning (CEL), a learning paradigm in which deployed robots share their execution experience to improve a common policy.

  Experience generated during deployment offers an opportunity for robots to improve policy reliability through learning from their own executions.

• [2]Learning from Fleet Experience: A Collective Learning Paradigm for Robots
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 03:41 AM · same content

Extractive summary: sentences quoted from the sources.