Surface Localization Using Camera-Detected Marks Instead of Gratings

The patent replaces the complex physical grating structure with a simplified digital mark pattern that is photographed and processed computationally. Instead of relying on physically manufactured gratings with high precision requirements, the system uses captured images of marks and applies image processing algorithms to achieve localization, effectively copying the measurement function from the physical domain to the digital domain.

The patent substitutes the mechanical grating measurement system with an optical imaging and computational processing system. Rather than using physical gratings that require precise mechanical manufacturing and alignment, the system captures images of marks and uses semi-supervised learning algorithms to extract position information, replacing mechanical precision requirements with computational processing.

The patent extracts the essential measurement function from the complex encoder system and isolates it to simple mark patterns that can be photographed by a camera. By taking out only the necessary elements (marks with specific patterns) and removing unnecessary components (complex gratings, encoders, optical systems), the system achieves high-precision localization with significantly reduced device complexity.

The patent replaces expensive, complex encoder components with simple, inexpensive mark patterns that can be easily manufactured and replaced. The marks serve as disposable or reusable simple features that don't require precision manufacturing, allowing the system to achieve high precision through computational methods rather than expensive hardware.

The patent introduces a dynamic, adaptive mark detection system using semi-supervised learning that can automatically adjust to different mark patterns and imaging conditions. Rather than relying on fixed, simple detection algorithms, the system uses machine learning models that are trained on labeled data and can adaptively improve detection accuracy, allowing simple mark patterns to yield high measurement precision through intelligent processing.

The patent changes the processing parameters and methods by applying semi-supervised learning algorithms that can handle variations in mark appearance, lighting conditions, and image quality. By transforming the detection approach from traditional fixed-threshold methods to adaptive machine learning-based methods, the system maintains simplicity in mark design while achieving high measurement precision through parameter optimization and intelligent processing.