Battery Product Visual Positioning Technology: The "Eagle Eye" of the Smart Manufacturing Era

In today's booming new energy vehicle and energy storage industries, batteries, as the core power source, directly determine product performance, safety, and lifespan through their manufacturing precision. Faced with increasingly stringent quality requirements and the demands of large-scale production, traditional manual inspection and mechanical positioning methods are no longer sufficient to meet micron-level precision manufacturing standards. Machine vision positioning technology, with its advantages of high precision, non-contact operation, and high efficiency, is becoming an indispensable "eagle eye" in the battery manufacturing field, reshaping the manufacturing precision of the entire process from cell to module to battery pack.

I. Overview and Core Value of Visual Positioning Technology

Visual positioning technology, in essence, utilizes industrial cameras, optical systems, image processing algorithms, and control systems to simulate and surpass the visual capabilities of the human eye, achieving rapid identification, precise positioning, and attitude measurement of target objects. In battery manufacturing, its core value is reflected in three aspects:

1. Precision Revolution: Improving positioning accuracy from the millimeter level of traditional mechanical methods to sub-millimeter or even micrometer level (e.g., ±0.05mm), eliminating performance degradation and safety hazards caused by assembly deviations at the source.

2. Efficiency Leap: Achieving 24/7 uninterrupted high-speed inspection, reducing single positioning time to less than 80 milliseconds, perfectly matching the high-speed production line cycle of tens or even hundreds of cells per minute, significantly improving production efficiency.

3. Quality Closed Loop: Achieving 100% online full inspection through a closed-loop control of "sensing-analysis-execution-feedback," effectively preventing human errors such as missing screws, incorrect sequence, and reversed polarity, intercepting defects before value-added processes, and driving manufacturing quality from one part per million (PPM) to one part per billion (PPB) – the ultimate manufacturing goal.

II. Application Scenarios of Visual Positioning in the Entire Battery Manufacturing Process

Visual positioning technology has deeply penetrated the entire "front-end, middle-end, and back-end" process of battery manufacturing, covering the four major manufacturing stages: electrode sheet, cell, module, and battery pack.

1. Front-end Electrode Sheet and Cell Manufacturing Stages

• Electrode Sheet Positioning and Correction: After coating, rolling, and slitting, the vision system accurately positions the electrode sheet edges and coating areas, guiding subsequent laser cleaning and busbar welding to ensure coating alignment and prevent active material misalignment.

• Winding/Layer Alignment: During the winding or layering process, the vision system monitors the relative positions of the separator and the positive and negative electrodes in real time, ensuring interlayer alignment accuracy (typically requiring <50μm) and preventing internal short circuits.

• Tab Welding Positioning: Guides the laser welding gun or ultrasonic welding head to accurately position it at the tab welding point, and allows for online inspection of weld quality (such as cold welds, missing welds, and weld misalignment) after welding.

2. Back-end Module and Pack Assembly

• Cell Loading and Arrangement Positioning: After AGV carts or conveyor lines transport the cells to the workstation, 2D/3D vision recognition identifies the precise position and orientation (X, Y, Z, θ) of the cells in the tray, guiding the robot for precise grasping and arrangement. This is compatible with cells of various sizes and models, solving the problem of inconsistent incoming material positions.

• Bolt Tightening Sequence Prevention: In battery pack assembly, an infrared vision positioning system such as Nexonar is used. By tracking infrared tags on the tool, the system identifies the positional relationship between the tightening gun head and the bolt holes in real time. The system only unlocks the tool to allow tightening when the gun head is aligned with the correct bolt and the sequence is correct, 100% preventing missed or incorrect tightening.

• Busbar Welding and Laser Cleaning Guidance: The vision system first accurately identifies the position of the cell terminals, then guides the laser head for surface cleaning (removing the oxide layer), and then guides the laser welding machine to weld the busbar to the terminals, ensuring accurate positioning and reliable connection of each weld point.

• Pack Appearance and Dimension Inspection: A full-view visual inspection is performed on the battery pack casing to check for gaps, flatness, scratches, bulges, etc., ensuring product appearance consistency and assembly quality to meet the stringent standards of vehicle manufacturers.

III. Key Technological Components of the Visual Positioning System

A complete battery visual positioning system is the product of deep integration of optics, mechanics, electronics, computing, and software.

1. Imaging Unit:

Industrial Camera: Based on accuracy and speed requirements, a high-resolution area scan camera or a high-frame-rate line scan camera is selected, and a global shutter is used to avoid motion blur.

Optical Lens: Lenses with appropriate focal length and depth of field are selected to ensure clear imaging.

Light Source System: This is crucial for success. Illumination schemes such as ring light, coaxial light, strip light, or structured light must be customized based on the reflective characteristics of the battery surface material (aluminum shell, copper terminals, blue film, etc.), and programmable brightness control must be implemented to highlight positioning features.

2. Processing and Control Unit:

Industrial PC/Vision Controller: Equipped with a real-time operating system, running professional machine vision software (such as Halcon, VisionPro) or deep learning algorithm platforms.

Core Algorithms: Includes image preprocessing (denoising, enhancement), feature extraction (edge, corner, blob analysis), template matching, coordinate transformation, etc. Sub-pixel-level positioning algorithms can improve recognition accuracy to less than one-tenth of the pixel.

3. Execution and Feedback Unit:

Robot/Servo Mechanism: Receives compensated coordinates sent by the vision system and performs actions such as grasping, placing, and welding.

PLC and Communication Network: Achieves high-speed, stable communication between the vision system, robot, and production line PLC through industrial buses such as EtherCAT and PROFINET, forming a real-time closed-loop control.

4. Technological Evolution: From 2D to 3D, from Traditional Algorithms to AI Integration

• 2D Vision: Suitable for planar positioning with distinct features, such as electrode edges and QR code reading.

• 3D Vision: Employing structured light, laser profilometers, or binocular stereo vision, it directly acquires the depth information of objects, offering irreplaceable advantages for detecting 3D features of battery casings such as pits, protrusions, weld height, and seal flatness, with detection accuracy reaching 0.1mm or even higher.

• AI and Deep Learning: Addressing the challenges of diverse types, varied shapes, and complex backgrounds of battery surface defects. By training deep learning models (such as CNN), the system can automatically learn defect features, achieving high-accuracy identification and classification of small targets and irregular defects, significantly reducing over-detection and under-detection rates. CATL's "Library of Lithium-ion Battery AI Large Model Data Base" further upgrades process development from an experience-based "trial and error" approach to a data-driven "predictive" approach.

IV. Typical Application Cases and Results

Case 1: Intelligent Tightening and Error Prevention of Bolts on Power Battery Pack Lines A leading battery manufacturer deployed TuYang Technology's ILS 3D Vision Positioning Error Prevention System on its Pack assembly line. The system uses a 3D camera to track the marking plate on the electric arc gun, comparing the position of the gun head with hundreds of bolt holes in real time. After implementation, it successfully reduced the risk of missing or incorrectly tightened bolts due to human error to zero, significantly improved the first-pass yield of welding and assembly, and markedly improved the overall equipment efficiency (OEE) of the production line.

Case Study 2: Fully Automated Sorting and Loading of Cylindrical Battery Cells

On the OCV (Open Circuit Voltage) testing and sorting line, Vision Dragon Technology used the VD200 system to solve the problem of loading mixed cells of various models. Camera at station 1 scans the tray, identifying the presence and position of all cells, optimizing the robot's pick-and-place path, and avoiding grabbing empty spaces; camera at station 2 corrects the deviation of cells on the suction cups, ensuring they are placed into the testing station with extremely high repeatability, guaranteeing the accuracy of voltage and internal resistance tests.

Case Study 3: AI-Powered Quality Inspection of Blue Film Appearance in Square-Shell Batteries

Addressing the industry pain point of difficulty in detecting appearance defects (bubbles, scratches, wrinkles) in batteries after they are coated with blue film, companies like Yihong Intelligent have launched a hexahedral inspection device integrating 3D vision and AI algorithms. A 3D camera acquires surface depth maps, and an AI model accurately distinguishes between actual defects and the inherent texture of the film material, achieving reliable detection of bubbles larger than 2mm in diameter and scratches wider than 2mm, ensuring the appearance quality of batteries leaving the factory.

V. Development Trends and Prospects

In the future, visual positioning technology for battery products will develop in the following directions:

1. Higher Precision and Speed: With advancements in camera sensor technology and processing chips, positioning accuracy will explore the nanometer level, while simultaneously meeting the higher production cycle requirements of next-generation battery production lines.

2. Enhanced Intelligence and Adaptability: AI will no longer be limited to defect classification but will be used for process parameter optimization, predictive maintenance, and real-time production line tuning. The system can learn on its own, adapting to rapid new product changes and achieving truly flexible manufacturing.

3. Multi-technology Integration and Digital Twins: Vision systems will integrate with more sensing technologies such as force sensors and infrared thermal imaging to provide more comprehensive quality data. Combined with digital twin technology, the entire positioning and welding process can be simulated and optimized in the virtual world and then mapped to the physical production line, significantly shortening the debugging cycle.

4. Standardization and Platformization: Leading manufacturers and alliances are committed to promoting the standardization of visual inspection interfaces and data formats, and building a unified industrial AI quality inspection platform to reduce integration complexity and cost.

Conclusion

From the micron-level alignment of electrode sheets to the error-free tightening of hundreds of bolts in pack assembly, visual positioning technology has permeated every aspect of intelligent battery manufacturing. It is not only a tool for improving accuracy and efficiency, but also the core of building a quality closed loop and realizing data-driven decision-making. With the arrival of the TWh manufacturing era, visual positioning, as a key enabling technology for the battery industry to move towards extreme manufacturing, will undoubtedly continue to deepen its application, safeguarding the manufacture of safer, more efficient, and more reliable battery products, and contributing to the grand blueprint of global energy transformation.