Dynamic Image Inspection Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In traditional assembly lines, image detection devices often operate at suboptimal levels due to lack of load balance, where one device being busy leaves others idle, leading to inefficient resource utilization.
Innovation Solution
A method that integrates the scheduling of GPU resources and distribution of inspection personnel by assigning images to multiple inspecting units, including both rule-based and model-based servers, and human inspectors, to dynamically allocate tasks and maintain load balance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If one image detection device is assigned to detect one specific feature, then the detection accuracy for that feature is improved, but other image detection devices become idle leading to low overall performance
Solution Approach 1:
The patent makes image detection devices universal by enabling them to detect multiple features rather than being dedicated to a single feature. The assigner dynamically assigns different features to be detected to different image detection devices based on their current status, allowing any device to perform any detection task. This multi-functionality resolves the contradiction by ensuring all devices remain productive while maintaining detection accuracy through appropriate task assignment.
Solution Approach 2:
The patent introduces dynamic task assignment where the allocation of detection features to image detection devices changes based on real-time device status. The assigner continuously monitors which devices are idle or busy and redistributes detection tasks accordingly. This dynamic approach ensures that detection accuracy is maintained for each feature while preventing device idle time, thus improving overall productivity.
2Reliability
If image detection devices are dedicated to specific features, then the detection reliability for each feature is improved, but load balance cannot be achieved leading to resource underutilization
Solution Approach 1:
By making image detection devices universal capable of detecting multiple features, the system achieves both reliability and adaptability. Each device can reliably perform its assigned detection task while also being adaptable to take on different tasks as needed for load balancing. The assigner manages this by tracking device capabilities and current status to optimize task distribution.
Solution Approach 2:
The patent implements a feedback mechanism where the assigner continuously monitors the status of image detection devices and adjusts task assignments accordingly. This feedback loop ensures that detection reliability is maintained while dynamically achieving load balance. The system responds to real-time conditions by redistributing detection features to prevent any single device from being overloaded while keeping others productive.
3Productivity
If multiple image detection devices work in parallel, then the productivity is improved, but without proper coordination devices become idle causing resource waste
Solution Approach 1:
The patent introduces an assigner as an intermediary that coordinates between multiple image detection devices and detection features. The assigner receives detection features, determines which devices are available, and assigns tasks to maintain balanced workload distribution. This intermediary prevents resource waste by ensuring that when multiple devices work in parallel, all of them remain productive without idle time, thus improving throughput while minimizing energy loss from underutilization.
Solution Approach 2:
The dynamic task assignment mechanism allows the system to adaptively coordinate multiple image detection devices working in parallel. The assigner continuously monitors device status and redistributes detection features to maintain optimal utilization. This dynamic coordination ensures high productivity through parallel processing while preventing resource waste by keeping all devices actively engaged in detection tasks.
Data Source
AI summary
A method for manufacturing a product is provided. The method includes: providing a plurality of components; working on the plurality of components to form the product; and testing the product via a testing method, including: providing first images and second images; assigning the first images by an assigner to a first inspecting unit; determining whether the product corresponding to the first images is OK or not in the first inspecting unit; additionally assigning the second image by the assigner to the first inspecting unit in case no more first image is to be assigned; determining whether the product corresponding to the second image is OK or not in the second inspecting unit or additionally in the first inspecting unit; and sending a message of OK by the assigner for outputting the product that is determined OK.