Robotic Arm Platform for 3D Reconstruction in Surgery

This technical report presents a robotic arm platform designed for multi-view image acquisition and accurate 3D reconstruction in minimally invasive surgery (MIS). The platform integrates a laparoscope with a robotic arm to capture images of various ovine organs under different lighting conditions and trajectories. The study employs advanced learning-based feature matchers in conjunction with COLMAP for generating 3D reconstructions, which are quantitatively evaluated against high-precision laser scans. The results indicate that while the reconstructions face challenges under realistic MIS conditions, certain configurations achieve sub-millimeter accuracy. The report outlines the importance of accurate 3D reconstructions for enhancing surgical navigation and planning, addressing the limitations of current datasets in the surgical domain. It also discusses the design and capabilities of the robotic arm, imaging systems, and the methodologies used for data acquisition and processing. The findings aim to contribute to improved surgical outcomes through enhanced visualization and guidance tools.