Checkout Data Reader Exception Handling with Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data reading systems face challenges in accurately reading optical codes due to errors caused by obscured barcodes, label quality issues, and specular reflections, leading to decreased reliability and increased operator intervention.
Innovation Solution
An automated checkout system with integrated exception handling and feedback mechanisms, including object recognition models and user verification, that captures images of items, decodes optical codes, identifies exceptions, and adapts object recognition models based on operator feedback to enhance data reading accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated exception handling is implemented, then productivity is improved, but device complexity increases
Solution Approach 1:
The system implements feedback mechanisms where operator resolutions of exceptions are captured and used to retrain object recognition models. The feedback loop includes: (1) capturing operator actions when they manually resolve exceptions, (2) storing these resolutions in a database, (3) periodically retrieving and analyzing the stored feedback data, and (4) retraining the object recognition models using this feedback to improve future automated exception handling.
Solution Approach 2:
The system enables self-service through automated exception handling where the object recognition models autonomously identify and resolve reading errors without operator intervention. The models automatically analyze optical code images, detect exceptions such as obscured barcodes or damaged labels, and attempt corrections using trained patterns from feedback data, reducing the need for manual operator involvement.
2Reliability
If operator feedback is collected and analyzed, then reliability is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary actions by continuously training and updating object recognition models in the background using accumulated operator feedback. This preliminary training occurs during low-traffic periods or asynchronously, so that when reading operations occur, the models are already optimized from previous feedback, minimizing the time impact on active reading operations.
Solution Approach 2:
The system implements periodic action by scheduling feedback analysis and model retraining at specific intervals rather than continuously during operations. The process periodically retrieves stored feedback data, analyzes it to identify patterns in operator corrections, and retrains models during designated training cycles, balancing reliability improvement with operational time constraints.
Data Source
AI summary
A checkout system for data reading, and related methods of use, the checkout system including one or more data reading devices with a conveyor for transporting items toward a read zone of the data reading devices, and an exception identification system capable of identifying exception items transported through the read zone without being successfully identified by the data reader. The checkout system includes an exception handling system operable to receive exception handling input for resolving an exception associated with the exception item, and a feedback system for receiving the exception handling input and determining whether and how to adjust an object recognition model of the data reading devices to improve performance.