Assembly Verification
Ensuring Perfect Component Assembly and Positioning
Overview.ai's Vision System is critical for ensuring components are correctly assembled, seated, and positioned, preventing costly errors and ensuring product functionality.
Key Applications
Soft Seated Connectors Inspection
We help automotive OEMs solve the long-standing problem of identifying loosely fitted or "soft seated" connectors, where a component is not fully or properly seated or engaged. It overcomes the complexities of inspecting these subtle defects on a continuously moving assembly line by using an advanced software auto-trigger mechanism for precise image capture. Its unparalleled AI algorithms can process and analyze hundreds of inspection points concurrently, even at different angles, ensuring comprehensive coverage and high-confidence checks to prevent electrical failures, fluid leaks, and safety concerns.
Speaker Panel Gap Detection
For automotive suppliers manufacturing speaker panels, the OV20i identifies gaps that prevent a snug fit in the corner, which is crucial for good performance. The system can identify correct parts, defect parts, and the location of defects for fixing, even though the mesh panels make identification difficult and gaps can appear in multiple locations. Overview AI achieved 97% accuracy with only 6 training images and in less than 30 minutes.
Black Clips in Foam Molds Presence/Placement
A major automotive supplier uses Overview AI to detect the presence and correct placement of black clips in foam molds. This challenge is amplified by eight different types of molds and altering mold appearance due to resin build-up. The OV20i implemented a [class/seg] recipe, achieving 100% accuracy in presence/absence tests with a handful of samples, and with additional training, detected incorrectly seated clips. The system proved robust to vibrations and resin build-up that had caused prior systems to fail.
Half Shaft Soft-Set Detection
For a major automotive OEM, the OV20i tackles the challenging inspection of detecting a soft-set condition in half shafts. These parts are often located deep inside the vehicle and partially obstructed, making them difficult to access visually. Overview AI trained a [class/seg] recipe with minimal training data to quickly assess pass/fail scenarios, identify correctly fit and soft-set conditions, and incorporate thresholds with additional training to reduce overkill and allow for part correction.
Loose Bolts in Engine Kits
The OV20i provides a solution for detecting loose bolts in engine kits, despite challenges like varying kit colors, shiny/oily finishes, and bolts often being hidden behind edges or varying significantly in size and orientation. Overview AI quickly trained and tested a segmentation recipe that could identify bolts in obstructed views, avoid reflections, and deal with multiple finishes, correctly identifying them in 47 out of 50 test cases with over 50 samples.
Seat Belt Cover Presence
For automotive suppliers manufacturing seats and belts, the OV20i implements a classifier recipe to identify the presence or absence of seat belt covers, addressing issues that led to A-ranked OEM defects. This inspection is particularly challenging due to black-on-black detection, variable lighting and placement, and the wide angle required. Leveraging its aligner tool to handle shifts and rotations, the system achieved 100% accuracy across all samples with only 10 images per component and 15 minutes setup time.
Seatbelt Guide Gap Length Measurement
The OV20i assists automotive suppliers with customer inspections for the gap in seatbelt guides, which previously led to escapes and chargebacks due to existing vision systems struggling with the required precision. Overview AI trained a classifier recipe that can directly measure the gap size and classify a pass/fail. With just 7 images and less than 30 minutes of training, the system accurately measured the gap, followed by Node-Red logic to define a clear passing threshold for high performance.
Breaker Base Inspection (Pitting, Flashing Lights, Inserts)
For a major electrical components manufacturer, Overview AI implemented a segmentation recipe to consolidate manual inspection of Breaker Bases. The solution addressed challenges like varied inspection needs, low contrast parts, and subtle pieces, allowing for the confirmation of flashing lights, detection of pitting, and inspection of small inserts within a single image. Initial issues were resolved through additional imaging and training to further reduce overkill.
Label Placement Verification (Medical Electronics)
Building on its success in contaminant detection, Overview AI extended a recipe for Valtronic to include identifying the correct placement of labels, allowing a single camera and program to manage both inspection issues. This helps medical electronics manufacturers ensure proper labeling compliance and reduce quality control overhead.