Analysis of a 2D Vision System for assembly tolerance compensation

In the aerospace industry, high precision and strict quality standards demand robust solutions for assembly and testing. This study presents the development and evaluation of a 2D vision system designed to detect and compensate for panel displacements during helicopter avionics testing. Traditional manual inspection is time-consuming and error-prone, especially when minor misalignments can lead to mechanical failures. The proposed system uses classical computer vision tech- niques—specifically, a pose estimation algorithm based on the PnP method—to determine the six degrees of freedom of a panel’s pose from a single image. A sim- ulation environment in Blender was used to generate synthetic images with con- trolled displacements, allowing precise performance evaluation. Results demon- strate that the system can achieve sub-millimetric accuracy while maintaining runtime constraints suitable for industrial implementation. The study concludes with a discussion on the impact of camera resolution, template design, and future improvements for integration into automated test procedures.