DIAB faced some issues with DXF and CNC file inconsistencies, such as missing measurements, misaligned parts, and discrepancies between 2D and 3D designs. These issues could affect production accuracy, leading to costly adjustments and delays. The primary objectives were to automate error detection, enhance the accuracy of cutting and nesting layouts, and streamline CNC file management to reduce manual inspections, improve material efficiency.

Project scope and our approach

AAI Labs took a systematic approach to tackle customer’s challenges, starting with an in-depth analysis of the DXF and CNC files. Here, recurring issues such as missing measurements, parts that did not match in assembly, and misaligned objects across different layers within the files were identified​.

Using the Python-based library, we mapped drawing objects to part numbers, allowing the system to detect six key error types, including missing measurements, detached dimension annotations, and discrepancies between equivalent part drawings. Each error type was categorized and systematically flagged, enabling to quickly identify and address issues that could disrupt production​

We implemented a method to analyze the polygons within each part entity, calculating overlap with reference polygons to assess part accuracy. This allowed the team to identify missing or mismatched parts by evaluating alignment and completeness across multiple file layers, which is essential for CNC machining precision​.

To improve material efficiency, we developed algorithms to optimize cutting and nesting layouts. These algorithms adjusted part orientations within the 2D layouts to achieve the best possible fit, reducing material waste and production costs. This step also included testing different optimization algorithms, such as graph-based and Monte Carlo methods, to ensure the best possible configuration for CNC and CAD processes.

Technological innovation and challenges

Our solution stands out for its ability to automate complex tasks within the CAD environment, bringing precision to a traditionally manual field. The main technological innovations included AI-enhanced error detection, where AI models detect common issues like missing measurements or misaligned parts in DXF files, allowing us to deliver a solution that proactively identifies and flags errors that would otherwise require manual checks.

Layer-specific analysis was also a significant innovation; advanced techniques enabled the team to differentiate between various file layers (e.g., “UNITS,” “CUTTING LINE,” “NESTING LAYOUT”) within the DXF files, accurately identifying and isolating inconsistencies across layers. This capability ensures strict quality standards by detecting even subtle discrepancies in part alignment and completeness. 

Additionally, comparative analysis using AI-based models allowed us to utilize a blend of machine learning and algorithmic analysis to compare “good” and “bad” file examples, identifying specific patterns that refined their error-detection algorithms for more accurate assessments and recommendations. 

However, we faced one key issue - handling inconsistencies within part representations across file layers, particularly when parts were rotated or had minor design variations. We overcame this by applying polygon rotation techniques to maximize overlap accuracy, providing a robust solution adaptable to different file orientations and configurations..

Project outcomes and future potential

Our solution enables streamlining the CAD review process significantly. The automated error detection and optimized nesting layouts led to substantial improvements in production accuracy and material usage efficiency.