How Scanflow Delivers AI-Based Orientation Inspection for Bonfiglioli

In addition to capturing label information, Bonfiglioli needed to automate the visual inspection of gearmotor orientation, specifically checking whether required components such as the clamp and bolt were correctly positioned. Manual visual inspection can be inconsistent and time-consuming, particularly when processing different gearmotor models.

Bonfiglioli engaged Scanflow to bring an AI-based visual inspection capability into the same Android application already used for label capture, checking clamp and bolt orientation immediately after the serial number scan.

1. No Automated Orientation Verification Existed There was no automated way to check whether components such as the clamp and bolt were correctly positioned on the gearmotor.
2. Manual Visual Inspection Was Inconsistent and Time-Consuming Checking orientation by eye left room for inconsistency between inspectors and added time to the inspection process.
3. Different Gearmotor Models Made Standardized Inspection Harder Processing different gearmotor models made it harder to apply a single, consistent inspection standard for orientation checks.

Scanflow proposed an AI-based Visual Inspection solution integrated into the Android application. The operator first scans the serial number and then captures an image of the gearmotor. Scanflow’s AI model performs visual inspection for Clamp and Bolt detection and returns a Matched / Not Matched result. If the result is not satisfactory, the operator can retry the capture; otherwise, they can proceed to the next scan. The solution supports 2 visual inspection orientations: Clamp and Bolt.

• Serial Number ScanThe operator scans the gearmotor’s serial number first, linking the orientation check to the same unit used for label capture.
• Gearmotor Image CaptureThe operator captures an image of the gearmotor using the Android device’s camera.
• AI Clamp & Bolt DetectionScanflow’s AI model inspects the captured image for correct clamp and bolt positioning.
• Matched / Not Matched ResultThe system returns an immediate Matched or Not Matched result to the operator at the point of inspection.
• Retry on FailureIf the result isn’t satisfactory, the operator can retry the capture; otherwise, they proceed to the next scan.
• Two Supported OrientationsThe solution supports two visual inspection orientations: Clamp and Bolt.

By adding AI-based orientation inspection to the same app used for label capture, Bonfiglioli gets a consistent, automated check on clamp and bolt positioning without introducing a separate workflow or device.

Key Outcomes

• Automates orientation checks — clamp and bolt positioning inspected by AI instead of manual visual review.
• Instant Matched/Not Matched result — the operator gets an output immediately after capture.
• Built-in retry — operators can recapture before moving on, rather than recording an uncertain result.
• Standardized across models — the same AI-based check applies regardless of gearmotor model.
• Result tied to serial number — every inspection outcome is recorded to Bonfiglioli’s database against the correct unit.

Deployment Footprint

• Android mobile/tablet camera for capture, integrated into the existing label capture workflow
• Server-side AI inference: AMD Ryzen 7 5800X + NVIDIA RTX 3060 12GB, running Ubuntu Server
• Matched/Not Matched results recorded to Bonfiglioli’s database through their provided API

Bonfiglioli strengthens gearmotor quality inspection with Scanflow, verifying clamp and bolt orientation automatically within the same workflow used for label capture.

Bonfiglioli now catches clamp and bolt orientation issues automatically, within the same app operators already use for label capture. Scanflow’s AI visual inspection returns an instant Matched/Not Matched result tied to the same serial number, with a built-in retry step before any result is recorded. The result is a consistent, standardized orientation check that removes the inconsistency of manual visual review without adding a separate device or workflow.