Automated Package Picking for Logistics Systems

Solution Overview

Solution Summary

Using a 3D machine-vision system, the position and pose of randomly stacked packages are detected, and the location data is streamed to a robotic arm. The robot then grasps the packages, replacing repetitive manual work and reducing package damage.

• Localization: Provides optimal grasping strategy and determines the pickup position.
• Classification: Distinguishes soft parcels from cardboard cartons to prevent parcel damage.
• Anomaly handling: Handles oversized, overweight, deformed, or stuck packages.

Vision System Components

• 1× FM851-E2 industrial 3D camera
• External LED light source
• Software SDK (industrial PC)
• Package Picking software package

Complete Solution Configuration

Camera:

• Camera hardware: Mech-Mind FM851-E2
• Vision software: Package Picking
• Robot control system: iNexBot control system

Industrial PC:

• OS: Windows 10 Enterprise 64-bit
• CPU: Intel(R) Core(TM) i7-10700F @ 2.90 GHz
• Memory: 16 GB
• GPU: NVIDIA GeForce GTX 1660 SUPER

Other: Light source, suction cup, chessboard calibration target, vacuum pump, etc.

Integrated System Performance

Lab condition: 1,680 pieces/hour Cycle time: 2.08 s/piece Real-world operation: > 1,500 pieces/hour

Advantage: Image acquisition and computation run in parallel with robot motion, so the robot never waits — zero idle time in the full workflow.

Vision System Installation

Camera installation height: 1,170 mm (measured from the top of the supply table)

Field of view:

• 725 × 635 @ 700 mm
• 1260 × 950 @ 1,170 mm

Vision System Performance

Built on the Mech-Mind active stereo structured-light 3D camera FM851-E2, the system combines a high frame rate with a large field of view, delivering an outstanding price-performance ratio for industrial-grade 3D vision.

Case Studies

Download

[Click here to start downloading the documentation](/api/cms-assets/downloads/手册中文版/3. 行业方案/物流供包行业方案/物流供包行业方案.pdf)