AI Vision Solution for Missing Step Detection in Industrial Assembly Process

In industrial production, the accuracy of assembly processes is of crucial importance. WeLinkirt (DaoAI) has solved the problem of missing step detection in assembly processes for a leading industrial SOP manufacturer with its advanced AI vision technology.

User scenario: A leading industrial SOP manufacturer has an assembly process on its production line that involves the assembly of various complex components. The product is high-precision industrial equipment. The detection object is whether each step in the assembly process is correctly executed, including the installation order of components, tightening degree, and whether the connection is in place. Due to the wide variety of products and complex assembly processes, it is difficult for manual inspection to ensure the accuracy and efficiency of detection.

Pain points: The manufacturer faces many quantitative dilemmas. First of all, the high yield rate of products leads to a natural scarcity of defect samples. At the same time, due to data confidentiality, it is difficult to collect a large amount of defect data for model training. This makes few-sample/single-sample modeling a key problem for the cold-start of detection. Secondly, manual inspection has a relatively high miss-detection rate and false-alarm rate. The miss-detection rate is about 5%, and the false-alarm rate is about 8%. This not only increases the defective rate of products but also leads to a large waste of labor costs in repeated inspections. In addition, when changing product models, the inspection process and standards need to be re-adjusted. The model-changing time is up to 30 minutes, which seriously affects production efficiency.

Technical principle

WeLinkirt uses advanced algorithms and imaging technologies to solve these problems. In terms of algorithms, the APDT positive-sample/few-sample learning algorithm in the DaoAI AI AOI software system is used. This algorithm can train the model with a small number of positive samples (1-20 good samples). The principle is to deeply mine and learn the features of good products and establish a feature-recognition model for the normal assembly process. When an abnormal situation is encountered, it can quickly identify the situation that does not match the normal features, thus realizing the missing-step detection. In terms of imaging, the self-developed 3D camera of the DaoAI 2D / 3D AI AOI device can perform 3D morphology reconstruction of the assembly process to obtain the accurate position and shape information of components. This is very effective for detecting hidden solder joints, coplanarity, and micron-level morphology because 3D imaging can provide more detailed information and is more accurate than traditional 2D imaging in judging whether the assembly is correct.

• The APDT positive-sample/few-sample learning algorithm can establish an efficient detection model by learning the features of a small number of good products, solving the problem of scarce defect samples.
• The 3D morphology reconstruction technology of the self-developed 3D camera can provide more accurate assembly information and improve the accuracy of detection.
• Based on the feature-recognition ability of the visual basic model, it can quickly identify abnormal situations in the assembly process.

WeLinkirt solution and product introduction

WeLinkirt provides a solution that combines the DaoAI AI AOI software system and the DaoAI 2D / 3D AI AOI device. The DaoAI AI AOI software system has strong feature-recognition ability. One good sample can complete 0-code automatic programming in only 5 minutes, greatly shortening the model-training time. Its APDT positive-sample/few-sample learning function can complete model training with 1-20 good samples, effectively solving the few-sample modeling problem. At the same time, the semantic false-alarm filtering function can reduce the false-alarm rate and improve the accuracy of detection. The self-developed 3D camera of the DaoAI 2D / 3D AI AOI device can comprehensively detect the assembly process. Its 3D morphology reconstruction technology can detect hidden solder joints, coplanarity, and micron-level morphology to ensure the correctness of the assembly process. During the implementation process, the device is deployed on the assembly production line and integrated through SDK / API / Docker, supporting 100% local private deployment to ensure that data does not leave the factory.

WeLinkirt's solution can effectively solve the problem of missing step detection in industrial assembly processes and improve production efficiency and product quality.

Quantitative results: By implementing WeLinkirt's solution, the manufacturer has achieved remarkable results. The detection rate has increased from the original 95% to 99.2%, and the miss-detection rate has been reduced to <0.8%, greatly improving the product quality. The false-alarm rate has been reduced by -65%, reducing the workload of manual re-inspection. The product model-changing time has been shortened from the original 30 minutes to 5 minutes, improving production efficiency.

This article was generated by AI. Customer cases are simulated scenarios based on real product capabilities and figures are illustrative; see product pages for official benchmarks.