Case Study | 3D Vision Enables Random Bin Picking & Feeding of Multi-Category Parts

In parts machining workshops, castings, sleeves and other workpieces are randomly stacked inside bins. Mutual occlusion and varied optical properties across different parts hinder conventional automation. Traditional manual sorting has long become a bottleneck restricting production line throughput.

Transfer Technology deploys a vision-guided picking workstation built around the Epic Eye Pixel Pro 3D industrial camera. It delivers precise position and pose estimation for multi-spec irregular workpieces, guides robotic arms to perform automated random feeding, and provides a stable, reliable automation solution.

On-Site Working Conditions

The project covers multiple component types including sleeves, tee castings and hollow brackets. Workpieces arrive in separate bins with random stacking inside. Severe optical challenges exist on site: black castings absorb light and complicate imaging; white tubular surfaces are prone to reflective interference. Overlapping and occluded workpieces further raise barriers to visual recognition.

Key pain points of the original production workflow:

◦ Heavy reliance on labor: Operators manually sort and orient parts. Long repetitive work leads to fatigue, with risks of mispicks and missed picks.

◦ Limitations of conventional vision systems: 2D vision lacks depth data. It cannot handle randomly stacked workpieces in bins and struggles to support multiple irregular component types.

◦ Throughput bottlenecks: Manual feeding creates cycle constraints and fails to match continuous operating demands of downstream machining equipment.

Customer transformation requirements: A single vision system shall handle multiple component types alternately without manual pre-arrangement of bin contents. The vision equipment outputs precise position and pose data to enable autonomous robotic picking and support long-duration nonstop production.

Customized Solution

The workstation adopts Transfer Technology’s Epic Eye Pixel Pro 3D industrial camera as the vision core, integrated with Yaskawa robotic arms and standard bins to build an all-in-one random bin picking station.

Anti-Interference 3D Imaging for Complex Optical Scenarios Targeting tough conditions including light-absorbing dark workpieces, reflective light-colored components and local occlusion, proprietary imaging algorithms optimize point cloud quality. The system effectively suppresses ambient light and surface reflection interference, fully reconstructing the 3D contour and spatial pose of workpieces. It addresses the industry-wide challenge of recognizing multi-material components at the source.

Multi-Part Compatibility & Intelligent Optimal Picking

A preloaded 3D model database works with AI matching algorithms to rapidly identify alternating incoming workpiece types. For stacked, tilted and interlaced parts in bins, the algorithm autonomously selects unobstructed targets and optimal grasping points, automatically avoiding stacking blind zones. Only a tiny fraction of severely interlocked, fully jammed workpieces remain unpickable. Automated picking covers most parts without manual tidying.

Standard Workflow

Material Loading: Operators deliver bins of workpieces to the station; no layering or pre-orientation is required.

3D Scanning & Reconstruction: The camera triggers scanning and generates complete scene point cloud data. AI Recognition & Motion Planning: The system identifies workpiece types and calculates optimal grasping positions and poses.

Robotic Picking & Transfer: The vision system transmits coordinate data; the robot picks parts and transfers them to target stations.

Cyclic Continuous Operation: After one bin is emptied, the system stands by automatically for the next batch.

Delivered Customer Value

◦ Stable production tact: Controllable automated picking cycles ensure seamless coordination of production line processes.

◦ Consistent machining quality: Standardized visual positioning eliminates workpiece collision damage and feeding misalignment caused by manual operation.

◦ Sustainable cost reduction & efficiency gains: Low maintenance requirements support long continuous operation, ideal for high-volume parts machining.

◦ Streamlined manpower allocation: Repetitive sorting and arranging tasks are automated to cut labor dependence and realize lean staffing at the workstation.

Random bin picking of machined components represents a widespread automation demand in the machining sector. Equipped with strong compatibility and anti-interference imaging performance, Transfer Technology’s Epic Eye series 3D industrial cameras are widely applicable for random bin picking of castings, sleeves, structural parts and other components.