Desarrollo de una plataforma mecánica de dos grados de libertad para la asistencia en la digitalización 3D de piezas mecánicas

Abstract

Three-dimensional scanning has gained relevance in fields such as reverse engineering, computer-aided design, and digital manufacturing; however, commercial equipment remains expensive and has limited accessibility in educational and technological institutions with restricted budgets in Ecuador. This thesis presents the development of an automated two-degree-of-freedom (pan-tilt) mechanical platform, built mainly from recycled aluminum obtained from beverage cans through sand casting, to assist in multi-angle image acquisition for low-cost photogrammetry processes. The general objective was to design and implement a mechanism capable of automatically controlling the position and orientation of a 16 Mpx Arducam IMX519 camera, integrated with a Raspberry Pi 4B with Pi-Shield, NEMA 17 stepper motors, and A4988 drivers, together with LED lighting and a polarizer to reduce reflections. Recyclable materials and low-cost components were used, achieving an estimated budget lower than that of comparable commercial scanners. The platform ensures stable trajectories, improving image overlap and enhancing the quality of the three-dimensional reconstruction through the open-source software OpenScanCloud (image acquisition followed by the generation of point clouds, meshes, and textures in .OBJ/.STL formats). Validation tests conducted with mechanical parts demonstrate a significant reduction in geometric errors and artifacts compared to scans performed using photogrammetry applications on mobile devices (such as Polycam, 3D Scanner App, or similar). This improvement is achieved through automated control of lighting intensity using adjustable LEDs within the acrylic enclosure, allowing the light level to be adapted according to the type of object and its surface characteristics. In this way, detail capture is optimized, reflections are minimized, and the overall quality of the reconstructed 3D models is improved. The results promote educational accessibility and replicability in low-resource environments.