Enabling High-Speed AOI Machines in Mail Sorting

Overview

AOI machines are a fundamental element in modern mail sorting environments, where envelopes and parcels move continuously and inspection must keep pace with the mechanical speed of the line. In high-speed AOI machines, visual verification, identification, and routing decisions must happen within a fixed timing window. Otherwise, even small delays or variability can directly affect sorting accuracy and operational efficiency.

For this reason, a leading mail sorting equipment manufacturer integrated Gidel’s FPGA-based vision infrastructure into its AOI machines. This integration enables a hardware-driven inspection pipeline that combines high-speed operation with reliable and accurate inspection results in production.

As a result, the system operates on a continuous sorting line, where items pass through the inspection zone without stopping or buffering. At the same time, multiple cameras capture image data simultaneously, covering different regions of each envelope. The system then delivers inspection results in time to trigger downstream sorting actions. This operational model places strict demands on synchronization, latency consistency, and sustained throughput.

The Challenge of High-Speed AOI Machines

Designing AOI machines for high-speed mail sorting presents several interconnected challenges that grow as throughput and camera count increase. The inspection system must process images from multiple cameras in parallel and operate continuously without performance degradation. At the same time, it must deliver results that align precisely with mechanical actuators and routing logic.

If the system relies too heavily on host CPU load or operating system scheduling, timing jitter can appear. Consequently, inspection reliability and sorting accuracy can degrade over time. To avoid this risk, the vision architecture must prioritize predictable behavior under sustained load.

• High-speed continuous motion: The system must inspect items on the fly without slowing the sorting line.
• Multi-camera inspection: Multiple cameras must cover different inspection regions in a single pass.
• Deterministic timing: The system must deliver inspection and routing decisions within a fixed latency budget.
• Continuous operation: AOI machines must run reliably for long periods under sustained throughput.

Vision Infrastructure Inside the AOI Machines

To address these challenges, the AOI machines rely on a hardware-centric vision architecture. In this design, FPGA-based frame grabbers handle image acquisition and preprocessing directly. This approach significantly reduces dependence on host CPU resources and operating system scheduling.

As a result, the system maintains consistent acquisition timing and stable data delivery even during peak throughput. At the same time, the architecture preserves inspection accuracy under continuous operation.

The customer used the HawkEye-CL Camera Link frame grabber as the core acquisition platform for multiple Camera Link cameras installed in the AOI machines. By anchoring acquisition on the FPGA, the system delivers deterministic frame timing, reliable camera synchronization, and stable data transfer into the processing pipeline, providing clean, consistent images that maximize the accuracy of downstream OCR and barcode reading algorithms.

In addition, this implementation reflects Gidel’s broader portfolio of FPGA Frame Grabbers, which target high-bandwidth, low-latency, multi-camera AOI applications where predictable behavior and inspection quality are equally important.

What the Frame Grabber Enables in AOI Machines

At the system level, the frame grabber serves as both a data capture device and a timing anchor for the inspection process. By performing acquisition and preprocessing close to the cameras, the AOI machines maintain consistent behavior regardless of downstream processing load.

• Deterministic multi-camera image acquisition
• FPGA-based preprocessing close to the sensor
• Stable latency under sustained inspection load
• Reliable synchronization across inspection zones

Managing Data Volume with FPGA-Based Compression

High-speed AOI machines generate large volumes of image data, especially during continuous multi-camera operation. Without data reduction, bandwidth and storage can quickly become system bottlenecks.

To prevent this, the OEM leveraged FPGA-based image compression directly on the frame grabber. This approach reduces downstream data rates while preserving image quality required for inspection, verification, and classification tasks. As a result, the system maintains predictable behavior even at high line speeds.

Embedded Processing with Mini Jetson Frame Grabber Systems

In AOI machine designs that require compact, embedded processing close to the inspection point, the architecture aligns naturally with Gidel’s Mini Jetson Frame Grabber Systems. These platforms combine FPGA-based image acquisition with NVIDIA Jetson processing.

Consequently, OEMs can integrate inspection logic, AI inference, recording, and control applications in a single platform. At the same time, the system preserves a deterministic hardware foundation that supports both speed and accuracy.

Operational Impact on High-Speed AOI Machines

By integrating FPGA-based vision infrastructure into the AOI machines, the OEM achieved stable throughput during continuous high-speed operation. The system also delivers predictable inspection latency that supports accurate routing decisions. In addition, FPGA offload reduces sensitivity to CPU load and operating system variability.

Together, these improvements translate into higher system reliability and smoother long-term operation on the sorting line.

• Stable throughput under continuous operation
• Consistent inspection latency for accurate sorting
• Reduced CPU load through FPGA offload
• Improved operational stability and inspection reliability

Conclusion

This case study demonstrates how Gidel’s technology enables high-speed AOI machines while maintaining reliable, real-time, and accurate inspection. By anchoring the vision pipeline on deterministic FPGA-based acquisition and preprocessing, and by integrating frame grabbers, compression technologies, and embedded vision platforms, the system achieves the combination of speed, accuracy, and long-term stability required in demanding mail sorting environments.

Revolutionizing Mail Sorting