Pallet Wrapper Computer Vision SKU Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The delivery process from distribution centers to stores is prone to errors and inefficiencies, particularly in accurately loading and unloading pallets of beverage products, leading to significant operational costs due to missing or mis-picked items, and lengthy wait times for drivers as clerks manually verify product quantities.
Innovation Solution
An improved delivery system utilizing machine learning and computer vision software, combined with RFID/Barcode technology, to validate pallet builds electronically, ensuring accurate SKU loading and counting before pallet wrapping, thereby preventing shortages and overages, and reducing verification time at the store by trusting the pre-loaded pallets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual verification by clerks is used to check pallet contents, then product accuracy can be confirmed, but driver wait time increases and operational efficiency decreases
Solution Approach 1:
The system performs preliminary verification of pallet contents using computer vision and machine learning models before the driver arrives at the store. The pallet is scanned and validated at the distribution center, ensuring product accuracy is confirmed in advance, thereby eliminating the need for time-consuming manual verification at the destination and reducing driver wait time.
Solution Approach 2:
The manual mechanical verification process performed by clerks is replaced with an automated optical system using cameras, computer vision algorithms, and machine learning models. The system captures images of the pallet, segments package faces, identifies SKUs, and validates contents automatically, substituting human labor with automated technological systems to improve efficiency.
2Reliability
If manual counting and verification of each case is performed, then product shortages and overages can be detected, but operational costs and time consumption increase significantly
Solution Approach 1:
The system creates digital copies of the physical pallet contents through image capture and processing. Instead of manually counting each case, the computer vision system generates virtual representations of packages, segments package faces from images, and uses machine learning to identify SKUs from these digital copies, enabling rapid verification without physical handling.
Solution Approach 2:
The system enables self-verification of pallet contents through automated scanning and validation. The pallet undergoes automatic inspection using cameras and AI models that independently verify product accuracy, count packages, and detect shortages or overages without requiring clerk intervention, thereby improving operational efficiency while maintaining verification reliability.
3Reliability
If extensive manual checking and pallet breaking down is required, then delivery accuracy can be ensured, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The system segments the verification process into distinct automated stages: image capture of the pallet, segmentation of individual package faces from the images, SKU identification from package faces using machine learning models, and validation against the pick list. This segmentation of the verification process into modular automated steps simplifies the overall complexity while ensuring delivery accuracy through systematic validation.
Data Source
AI summary
A delivery system may include a pallet wrapper system having a turntable, a camera directed toward an area above the turntable, and a stretch wrap dispenser adjacent the turntable. A computer receives images from the camera of multiple sides of a pallet loaded with packages on the turntable. The computer stitches images from different sides of the stack of packages that correspond to the same package. At least one machine learning model may be used to infer SKUs of each package. Optical character recognition may be performed in parallel on the images. The determination of the SKU of each package may be based upon the inferred SKUs and on the OCR.