← Back to the library
Schneider Electric picks warehouse items with an AI foundation model
Added 2026-10-14 · Reviewed 2026-10-15 · Schneider Electric
Read the source ↓What the company did
Geekplus reports installing its Robot Arm Picking Station at Schneider Electric's Shanghai warehouse. Geekplus Brain and zero-shot learning support item recognition and picking across varied SKUs, connecting the arm to existing mobile-robot and warehouse processes without per-item retraining, according to the provider.
How it went
For this pilot, Geekplus reports twice manual picking throughput against a 150–300 pieces-per-hour manual range, at least 99.99% accuracy and production readiness within 48 hours. The article does not publish the sample size or test duration; Freight Fieldnotes has not reproduced these claims.
Worth knowing
- The provider describes on-premises operation with data kept at the site. That architecture still depends on the installed warehouse equipment and software.
- For your desk: Compare picks across representative packaging and SKU changes, separating wrong-item picks, dropped items, damage and integration failures in the exception log.
Sources & provenance
Geek+ Wins 2026 RBR50 Innovation Award for Robot Arm Picking Station ↗
Geekplus · Published 2026-05-19 · Reviewed 2026-10-15
“successful deployment at Schneider Electric’s Shanghai warehouse”
Supporting evidence and review
Human review required
- Independent review advice: check pick identity, item damage and exception handling across representative SKUs.
- Validate integration failures separately from robotic grasp failures.
Evidence & limitations
- Company/provider-reported deployment; no independent performance evaluation by Freight Fieldnotes.
- The award report includes a named pilot deployment.
- Accuracy and deployment-speed figures are provider-reported; no independent test is reproduced here.