In the automotive manufacturing industry, the accuracy of assembly is crucial. DaoAI provides an efficient solution to the problem of missing and incorrect installation in assembly with advanced AI vision technology.
User Scenario: A leading automotive parts manufacturer's assembly line mainly produces automotive engine assemblies. The inspection objects focus on the assembly status of various parts during the engine assembly process, including small parts such as bolts, nuts, and gaskets, as well as key components such as sensors and controllers, to ensure the integrity and accuracy of the assembly.
Pain Points: Previously, the production line used traditional manual inspection and rule-based machine vision inspection methods. In terms of manual inspection, the labor cost was high, and after long-term work, inspectors were prone to fatigue, resulting in a missed detection rate as high as 3%. At the same time, the inspection cycle of manual inspection was long, with only one inspection completed every 180ms, seriously affecting production efficiency. Although the rule-based machine vision inspection could improve the inspection efficiency to a certain extent, in the face of complex assembly scenarios and unseen missing or incorrect installation situations, the false alarm rate was as high as 20%. Moreover, every time the production model was changed, it took 2 hours to reprogram and debug, seriously affecting the production progress and product quality.
Technical Principle
DaoAI adopts advanced unsupervised anomaly detection algorithms and 3D imaging technology. In terms of algorithms, the unsupervised anomaly detection only needs to use good products for modeling. It constructs a standard model by learning the feature distribution of good products. When the features of the inspection object deviate significantly from the standard model, it can be determined as an anomaly, that is, a missing or incorrect installation situation. The strength of this algorithm lies in its ability to recognize missing or incorrect installation defects that have never appeared in the training data, greatly improving the generalization ability of detection. In terms of imaging, the self-developed 3D camera combined with 3D morphology reconstruction technology can accurately obtain the 3D morphology information of the engine assembly. Compared with traditional 2D imaging, 3D imaging can capture the depth and shape information of objects, which has significant advantages in detecting hidden welds, coplanarity, and micron-level morphology, and can more accurately determine whether parts are assembled correctly.
- The unsupervised anomaly detection algorithm learns the feature distribution of good products, constructs a standard model, and identifies abnormal situations.
- The self-developed 3D camera obtains 3D morphology information and cooperates with 3D morphology reconstruction technology to improve detection accuracy.
- 3D imaging can capture depth and shape information and effectively detect hidden and subtle assembly problems.
DaoAI's Solutions and Product Introduction
DaoAI provides the DaoAI AI AOI software system and the DaoAI 2D / 3D AI AOI equipment. The DaoAI AI AOI software system has a powerful feature recognition ability of the visual basic model. It can complete zero-code automatic programming in 5 minutes with only one good product. The APDT positive-sample/few-sample learning method it uses can perform efficient learning with only 10 good products. At the same time, the semantic false alarm filtering function can effectively reduce the false alarm rate. The DaoAI 2D / 3D AI AOI equipment integrates the self-developed 3D camera and 3D morphology reconstruction technology, which can accurately detect problems such as hidden welds, coplanarity, and micron-level morphology. In the implementation, the equipment is installed at key positions on the assembly line. The software system processes the data collected by the camera in real-time, quickly and accurately determines whether there are missing or incorrect installation situations, and feeds back to the production line for adjustment in time.
DaoAI's products bring an efficient solution to the detection of missing and incorrect installation in automotive assembly with advanced technology.
Quantifiable Results: After adopting DaoAI's solution, the inspection effect has been significantly improved. The detection rate has reached 99%, and the missed detection rate has been reduced to <1%, greatly improving product quality. The false alarm rate has been reduced by -85%, reducing unnecessary manual re-inspection and downtime adjustment time. The production model change time has been shortened from the original 2 hours to 5min, greatly improving production flexibility and efficiency.