Learning from Fleet Experience: Collective Experience Learning Across 96 Robots
Learning from Fleet Experience: Collective Experience Learning Across 96 Robots
This research proposes Collective Experience Learning (CEL), a paradigm where deployed robots pool execution experience to update a shared policy. The method, implemented via Failure-Guided Experience Retrieval Fine-Tuning (FERT), uses failure contexts to retrieve successful trajectory segments for supervised updates without requiring new expert demonstrations. Evaluated across five real-world manipulation tasks on robots excluded from training, the approach achieves success rates between 93.3% and 99.2%. Specifically, on multi-stage phone packaging, using a fleet of 96 robots improved success rates from 79.2% to 98.3%, validating CEL as a scalable method for improving robot policy reliability through deployment-generated experience.
The paper introduces Collective Experience Learning (CEL) and Failure-Guided Experience Retrieval Fine-Tuning (FERT), demonstrating that policies trained on failure contexts from a fleet of 96 robots achieve 93.3-99.2% success rates across five real-world manipulation tasks without additional expert demonstrations.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Data origin
- Robot trajectoriesOpen source S3
Reported numbers
Reported success rate
93.3–99.2%
Reported trials
View original evidence
demonstrations or corrective actions. Operators provide binary episode-outcome labels and environment resets. A fault-tolerant asynchronous training system (FATS) coordinates robot execution, cloud inference and policy optimization despite local interruptions. Across five real-world manipulation tasks, the resulting policies achieve success rates of 93.3-99.2%, evaluated over 120 trials per task on robots excluded fr
Open source S4trials per task
>120 trials
Reported trials
View original evidence
demonstrations or corrective actions. Operators provide binary episode-outcome labels and environment resets. A fault-tolerant asynchronous training system (FATS) coordinates robot execution, cloud inference and policy optimization despite local interruptions. Across five real-world manipulation tasks, the resulting policies achieve success rates of 93.3-99.2%, evaluated over 120 trials per task on robots excluded fr
Open source S4robots in fleet
96 robots
View original evidence
Learning from Fleet Experience: Collective Experience Learning Across 96 Robots
Open source S1tasks evaluated
5
Reported trials
View original evidence
demonstrations or corrective actions. Operators provide binary episode-outcome labels and environment resets. A fault-tolerant asynchronous training system (FATS) coordinates robot execution, cloud inference and policy optimization despite local interruptions. Across five real-world manipulation tasks, the resulting policies achieve success rates of 93.3-99.2%, evaluated over 120 trials per task on robots excluded fr
Open source S4
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2610.25736v3 Announce Type: replace Abstract: Robot deployment generates experience that can support policy improvement, but continued reliance on expert corrective actions makes learning difficult to scale across growing fleets. We propose learning from fleet experience: using deployed robots' own executions as a source of supervision for further policy improvement. We call this paradigm Collective Experience
Open source S2
Learning (CEL), in which robots pool execution experience to update a shared policy that returns to the fleet. Following expert-supervised initialization, we realize CEL through failure-guided experience retrieval fine-tuning (FERT). FERT uses failure contexts to retrieve relevant segments from successful trajectories and prioritizes their recorded actions for supervised updates, without collecting additional expert
Open source S3
demonstrations or corrective actions. Operators provide binary episode-outcome labels and environment resets. A fault-tolerant asynchronous training system (FATS) coordinates robot execution, cloud inference and policy optimization despite local interruptions. Across five real-world manipulation tasks, the resulting policies achieve success rates of 93.3-99.2%, evaluated over 120 trials per task on robots excluded fr
Open source S4
om training data collection. With collection time per robot held fixed, larger fleets provide more experience and generally improve performance across the evaluated sizes. On multi-stage phone packaging, learning with 96 robots raises success from 79.2% to 98.3%. These results support CEL as a scalable approach to improving robot policy reliability through experience generated during deployment.
Open source S5
Learning from Fleet Experience: Collective Experience Learning Across 96 Robots
Open source S1
Implications for data suppliers
RoboSignal interpretation and collection questions, not statements of buyer demand.
- Match the buyer’s robot embodiment, action space, sensors and timestamp requirements before planning collection.
- Compare the reported units and scope before using these quantities in a budget. Recording hours, sensor-hours and trajectories are different measures.
- Validate demand and acceptance criteria with a buyer before scaling. Publication, popularity and a research result do not establish a purchase commitment.
Source：arXiv Robotics — research abstracts · arxiv.org