CASO DE ESTUDIO

Perfecting Battery Lid Epoxy Inspection

Descargar estudio de caso

Battery Lid Epoxy Inspection Case Study | UnitX

CONTENIDO

COMPARTIR TAMBIÉN

Introduction: The “Live Feed” Fatigue

In battery manufacturing, the epoxy seal on the lid is the final barrier against the elements. Any deviation in application—such as air bubbles, path gaps, or insufficient volume—compromises the seal’s integrity, posing severe leakage and safety risks.

Our client faced the complex challenge of identifying translucent epoxy defects across a range of surface finishes, where low visual contrast frequently compromises the reliability of traditional inspection.

Previously, they relied on a “Live Feed Camera”—a passive video monitoring system. This required operators to manually verify the epoxy dispense for every unit by watching a continuous screen feed in real-time. This approach was not only labor-intensive but highly susceptible to “inspection fatigue.” Over time, the subtle visual cues of a gap or bubble became easy to miss, leading to inconsistent quality standards. It became clear that the client needed to evolve from subjective manual observation to an autonomous, data-driven inspection solution.

The Challenge: Seeing the Clear on the Shiny

The project goal was to replace the human operator with an AI system that could quantitatively measure the epoxy application.

  • Métricas objetivo: False Acceptance (FA) < 1%, False Rejection (FR) < 5%.
  • Defect Targets: Identifying minute Burbujas de aire (voids) and Brechas (shortages/low level) in the epoxy bead.
  • Tiempo del ciclo: The inspection had to complete in <15 segundos to match the robotic dispensing speed.

Seeing the Clear on the Shiny

The complex geometry of the battery lid makes consistent epoxy application—and inspection—a challenge

The Solution: A Robotically Enabled UnitX Sistema de inspección

UnitX deployed a robust, robotically automated solution across two production lines.

Arquitectura del sistema:
To maintain high spatial resolution across the entire field of view, we utilized a high-spec optical setup:

  • The “Eye”: A Large FOV (Field of View) UnitX OptiX system paired with a 24 MP camera and panel light. This combination provides the necessary contrast to see transparent epoxy against plastic .
  • The “Brain”: A UnitX Corteza unit handles the processing locally, ensuring low latency.

A Robotic CorteX Edge Actualización

El proceso de UnitX inspection cart, integrated directly into the robotic dispensing line

Results: Achieving the “Impossible” 0/0 Split

The shift to an AI-driven architecture exceeded all projected KPIs. Formal validation confirmed that the system delivered a stabilized quality control framework alongside a measurable increase in production yield.

1.The “Zero-Error” Performance:

  • Zero Escapes (0% False Acceptance Rate): Absolute detection of every bubble and gap, ensuring that potential leakage risks are mitigated at the source before they can become costly warranty claims.
  • Optimización del rendimiento: This system achieves a 1.48% False Rejection Rate, drastically reducing the “overkill” common in traditional vision systems. By eliminating false flags, it preserves yield and prevents the unnecessary disposal of high-quality parts.

2. Despliegue Rápido  

  • Days to SAT: 12 Days per line.

From installation to Prueba de aceptación en el sitio (SAT) took less than two weeks. This rapid speed meant the client could realize the value of the system almost immediately.

3. Excelencia Operacional 

  • Elimination of Warranty Claims: By catching seal defects at the source, the client stopped shipping potential leakers.
  • Costos laborales reducidos: The operator previously assigned to watch the live feed was repurposed, and the inspection is now fully automated and conducted robotically.

Achieving the Impossible 0 Split

UnitX AI visualizes defects that are hard to spot manually, clearly marking air bubbles and epoxy gaps

Conclusión

Reliance on operators for “Live Feed” manual inspection is a bottleneck that modern manufacturers can no longer afford. This case study demonstrates that even for difficult materials like transparent epoxy on plastic, UnitX AI can achieve 100% detection accuracy. The result? Less scrap, zero warranty claims, and a safer final product.

Stop staring at live feeds. Start automating.
Contacto UnitX to upgrade your dispensing lines today.

Estudio de caso relacionado

imagen 91 (1)
Grupo 20
foto 56
foto 52
Grupo 9
Grupo 19 (1)
Grupo 6 (1)
Grupo 27 (1)
foto 59
foto 60
Ir al Inicio