WeLinkirt DaoAI 3D Robot Vision (proprietary 3D camera + 6D pose estimation, bin picking, glue dispensing/assembly/loading/unloading guidance, brain-eye-body closed-loop, sub-millimeter hand-eye coordination) automates the collection and analysis of critical quality data by performing high-precision 3D morphology inspection and defect identification on new energy battery module solder joints, thereby reducing the average time spent on manual quality traceability by −70%.
As the core of electric vehicles and energy storage systems, the performance and safety of new energy batteries directly depend on every detail in the manufacturing process. Among them, the cell connection solder joints within the module are critical factors affecting battery consistency, internal resistance, and lifespan. To ensure long-term product reliability, leading new energy battery manufacturers have extremely stringent requirements for solder joint quality inspection. However, traditional inspection methods often only detect overt defects, leaving hidden quality risks that could lead to early failure, and how to effectively integrate these inspection data into the production quality traceability system, as long-standing challenges in the industry. Especially in the face of increasing production scale and pressure to be responsible for the full lifecycle quality of products, an automated solution capable of achieving a closed data loop for solder joint quality is urgently needed.
Pain Points: Why This Hurdle Is Difficult to Overcome
A leading new energy battery manufacturer faced multiple challenges in module solder joint inspection. Firstly, the missed detection rate of manual visual inspection or traditional 2D AOI systems remained high, often above 1% for subtle defects such as micro-cracks, internal pores, and cold solder joints, leading to risks of early product failure. Secondly, due to the lack of an effective quality data traceability mechanism, once a batch-specific market issue was discovered, tracing it back to specific production batches, modules, or even solder joint processes and parameters was time-consuming, averaging 2–3 days, severely impacting problem response speed and customer satisfaction. Thirdly, traditional inspection solutions struggled to integrate data with subsequent production management systems (MES/SCADA), preventing real-time feedback of inspection results for upstream process adjustments, resulting in long quality improvement cycles and high false positive rates, leading to significant manual re-inspection workload, costing 4–6 hours daily and increasing operational costs. These accumulated pain points not only constrained capacity expansion but also increased compliance risks.
Delving into the root causes, the difficulty of inspecting solder joints in new energy battery modules lies in their complexity and diversity. On one hand, solder joints have irregular shapes and highly reflective molten metal surfaces, making traditional 2D vision highly susceptible to ambient light and surface textures, thus unable to accurately acquire 3D morphological information. On the other hand, defect types are diverse, ranging from macroscopic defects visible to the naked eye (e.g., misalignment, insufficient solder) to micron-level sub-surface defects (e.g., micro-cracks, pores), and these defects are often hidden within the solder joint, making them difficult to identify through surface features. Furthermore, with the iteration of battery technology, solder joint materials and processes are constantly changing, making traditional rule-based AOI systems costly to retool and difficult to adapt to multi-variety, small-batch production demands. These factors make precise, automated, and traceable inspection of module solder joint quality a formidable challenge. The current industry buzz around low-cost humanoid robot development platforms, with its core focus on the popularization of embodied AI technology, aligns perfectly with WeLinkirt DaoAI 3D Robot Vision's 'brain-eye-body' closed-loop concept, empowering machines with 'perception' and 'decision-making' capabilities for intelligent operation in complex industrial scenarios.
Technical Principles
WeLinkirt DaoAI 3D Robot Vision system, through its proprietary high-precision 3D camera and advanced 6D pose estimation technology, thoroughly solves the challenges of new energy battery module solder joint inspection. The system employs structured light or laser triangulation principles to perform non-contact scanning of solder joint surfaces, reconstructing precise 3D morphology with sub-millimeter accuracy. This allows even minute defects such as protrusions, depressions, cracks, or edge collapses to be clearly captured in 3D space. Unlike traditional 2D vision, which relies solely on grayscale or color information, DaoAI 3D Robot Vision acquires real geometric data, inherently robust to issues like glare and insufficient contrast. Furthermore, its powerful 6D pose estimation capability ensures that robotic arms can precisely locate each solder joint even in complex, unorganized environments, achieving high-precision, high-repeatability inspection.
Compared to traditional manual visual inspection, WeLinkirt DaoAI 3D Robot Vision system not only improves inspection speed and consistency but fundamentally resolves missed detections and false positives caused by human fatigue and subjective judgment. Compared to rule-based traditional AOI systems, DaoAI's AI-driven defect recognition algorithms, especially the APDT few-shot self-training capability, can quickly train high-precision defect models with only 1–20 good samples, significantly reducing changeover time and effectively identifying complex defects difficult for traditional algorithms. More importantly, the WeLinkirt system can upload critical parameters such as each solder joint's 3D data, defect type, and position information in real-time to the MES system, building a complete quality traceability chain and achieving a data closed-loop, providing strong support for subsequent process optimization and product recalls. For instance, in practical applications, the system can boost solder joint defect detection rates to over 99.5% while reducing the false positive rate by −85%.
Typical Application Scenarios
- **Solder Joint Height and Coplanarity Inspection**: Precisely measures the height of solder joints between cell connecting tabs and busbars within the module, ensuring good coplanarity to prevent cold solder or poor contact. The challenge lies in the multitude of closely spaced solder joints, making individual measurement difficult for traditional methods.
- **Solder Joint Surface Defect Detection**: Identifies micro-cracks, pores, shrinkage, oxidation, solder dross, cold solder, and other minute defects on the solder joint surface. These defects are often tiny and can be obscured by glare; DaoAI 3D Vision can penetrate surface gloss to acquire true geometric morphology.
- **Solder Joint Morphology Integrity and Consistency**: Evaluates whether solder joints are full and uniform, free from collapse, insufficient or excessive solder, ensuring each solder joint's morphology meets design standards and improves overall module electrical performance. The difficulty lies in complex standards and the need for global judgment of 3D morphology.
- **Solder Joint Position Deviation and Bridging Detection**: Precisely measures the offset of solder joints relative to their designed positions and detects the risk of short circuits due to bridging between adjacent solder joints. High-precision 6D pose estimation is key to achieving this detection.
- **Solder Joint Size and Volume Measurement**: Quantitatively measures key dimensions of solder joints such as diameter, length, and volume, ensuring solder joint strength and conductivity meet standards. These 3D dimensional information are inaccessible to traditional 2D vision.
Implementation Case Study
A Tier-1 supplier specializing in high-end electric vehicle battery module production had extremely high demands for solder joint quality traceability on its production line. Previously, the manufacturer used a combination of manual inspection and traditional 2D AOI, but still faced high missed detection rates and data silo issues. The WeLinkirt DaoAI 3D Robot Vision team provided a customized solution. In the initial phase, we thoroughly analyzed their production line characteristics and solder joint defect types. Using the DaoAI 3D camera, we performed multi-angle scanning of the module solder joints and, combined with APDT few-shot learning technology, quickly trained a defect model with only 15 good samples. After deployment, the system seamlessly integrated with the client's MES system, associating and uploading real-time 3D morphological data, defect types, position coordinates, and inspection results for each module and solder joint to the database.
Post-implementation, the client's production efficiency and quality management capabilities significantly improved. Before deployment, the average missed detection rate for module solder joint inspection was 0.9%, leading to significant additional human resources spent monthly on rework and traceability, and occasional early failure reports in the market due to solder joint issues. After deployment, through the precise inspection and data closed-loop of the WeLinkirt DaoAI 3D Robot Vision system, the solder joint missed detection rate was reduced to <0.4%, and the false positive rate was reduced by −80%. More critically, quality issue traceability, which previously took 2–3 days, can now be precisely located to specific modules and production batches within 1 hour, even tracing back to process parameters for specific time periods, greatly improving problem response speed and customer satisfaction. This solution saved the manufacturer significant labor costs and substantially enhanced product competitiveness in the market.
"DaoAI 3D Robot Vision not only improved the precision of our solder joint inspection but, more importantly, it built a complete quality traceability chain, giving every battery module a 'digital ID card' for its quality."
WeLinkirt Solution and Products
WeLinkirt's core solution for new energy battery module solder joint quality traceability and data closed-loop is centered around the DaoAI 3D Robot Vision system. Our proprietary high-precision 3D camera is the foundation for acquiring the 3D morphology of solder joints. Combined with a high-performance computing platform, it enables rapid scanning and data processing of thousands of solder joints per second. The system's core lies in its powerful AI algorithms, including a deep learning-based 6D pose estimation module, ensuring that the robotic arm can precisely guide the 3D camera to target solder joints, adapting autonomously even to slight module displacements. Defect recognition relies on the DaoAI AI AOI software system, whose built-in visual foundation models can learn complex features of solder joints and, combined with APDT few-shot self-training technology, quickly adapt to new solder joint processes and defect types, achieving 0-code rapid changeover, with new product inspection plan configuration completed in just 5 minutes.
The WeLinkirt DaoAI 3D Robot Vision system supports 100% on-premise private deployment, ensuring customer data security and meeting high-level data compliance requirements. We offer various integration methods such as SDK/API/Docker, facilitating seamless embedding into existing MES/SCADA systems for real-time interaction between inspection and production data. By establishing a 'brain-eye-body' closed-loop—visual (eye) perception of defects, AI (brain) decision analysis, and robot (body) execution of gripping or guidance—the system not only identifies defects but also correlates defect data with production process parameters, forming a data-driven quality improvement loop. This system can increase the overall OEE of the production line by over 15% and significantly reduce rework and scrap due to quality issues, creating substantial economic benefits and brand value for clients.
FAQ
How does DaoAI 3D Robot Vision achieve precise traceability of solder joint quality?
The DaoAI 3D Robot Vision system acquires 3D morphological data for each solder joint using its high-precision 3D camera, combined with AI algorithms to identify defect types and locations. These detailed inspection results are associated in real-time with production batches, module serial numbers, and other information, then uploaded to the MES system, building a complete quality data chain from raw materials to finished products, ensuring every solder joint has a traceable 'digital ID card'.
How adaptable is this system to solder joints of different shapes and materials?
DaoAI 3D Robot Vision boasts excellent adaptability. Its proprietary 3D camera obtains true 3D morphology, unaffected by surface glare or contrast. Coupled with APDT few-shot self-training technology, it requires only a small number of good samples to quickly train new defect models, enabling 0-code rapid changeover. This easily adapts to inspection needs for solder joints of different shapes, materials, and processes, reducing changeover costs and time.
How does WeLinkirt ensure the security and privacy of inspection data?
WeLinkirt DaoAI 3D Robot Vision system supports 100% on-premise private deployment. All inspection data is processed, stored, and managed within the client's factory, with no data uploaded to the cloud or leaving the client's network environment. This adheres to stringent data security and privacy protection requirements, providing clients with a highly secure and reliable solution.