Automating the “final picking” process long considered dependent on manual labor
Eureka Robotics (Headquarters: Singapore; Japan Office: Koto-ku, Tokyo; Managing Director Japan: Hironori Ishimaru; hereinafter “Eureka”) announced today that its Eureka AI Vision System has been selected as the standard vision solution for a universal automated picking system currently under development by logistics system integrator AIOI SYSTEMS Co., Ltd. (Headquarters: Shinagawa-ku, Tokyo; President: Yutaka Yoshino; hereinafter “AIOI SYSTEMS”).
With this adoption, AIOI SYSTEMS is building a solution framework capable of proposing automation across the entire logistics workflow—from upstream processes to downstream operations—including the final picking stage, which has historically been difficult to automate.
Both companies will continue conducting verification tests with the goal of practical deployment in logistics environments, further validating performance and enhancing the system for real-world operation.
Background: The “Last Barrier” in Logistics Automation
In logistics operations, automation investments have accelerated in recent years due to severe labor shortages. Processes such as receiving, storage, transport, and sorting are increasingly automated.
However, the final picking process before shipment remains challenging to automate. Logistics environments must handle a wide variety of products with constantly changing SKUs, and conventional robotic picking solutions typically require CAD models or prior data registration for each product. This creates a significant operational burden, leaving many facilities reliant on manual labor.
The Solution: Universal Automation That “Sees and Picks”
To address this challenge, AIOI SYSTEMS developed a universal automated picking system capable of picking virtually any product simply by “seeing” it, by adopting the Eureka AI Vision System, which serves as the robot arm’s “eyes and brain.”
Because the system does not depend on CAD models or prior object registration, it can recognize and pick products exactly as they appear in real operations. This enables reliable automation even in high-mix, constantly changing logistics environments.
System Overview: Masterless Picking
The system integrates an industrial robot arm with the Eureka AI Vision System, which consists of a Eureka 3D Camera and a Eureka Controller.
When the Eureka 3D Camera captures images of products, the Eureka Controller generates 3D point-cloud data in real time. Using this data, Eureka’s proprietary AI and algorithms automatically perform several key processes:
- High-speed 3D data generation
- Detection of pickable objects
- Estimation of optimal grasp points
- Calculation of robot pose (6DoF)
The robot arm then picks the object and places it in the designated location.
A key feature of the system is its masterless picking approach, which eliminates the need for CAD models or prior object registration. Traditional logistics picking systems require model registration, CAD preparation, and reconfiguration whenever new SKUs are added.
By directly recognizing real objects in 3D as they appear on the shop floor, the system enables flexible operation in high-mix environments.
This makes it particularly suitable for facilities such as e-commerce warehouses, where product assortments change frequently. The system enables continuous operation with minimal system stoppage or reconfiguration.
Why Eureka: AI Vision Designed for Logistics Reality
AIOI SYSTEMS selected the Eureka AI Vision System based on several features that align closely with real logistics environments:
- Picking without CAD models or prior data registration
- No training required, simplifying setup and operation
- Stable object detection using a masterless model
These capabilities enable automation systems that continue operating without interruption even in environments where SKUs change frequently.