MiDAS: A Multimodal Data Acquisition System and Dataset for Robot-Assisted Minimally Invasive Surgery

Researchers have released MiDAS, an open-source, platform-agnostic system that captures synchronized multimodal data from surgical robots without needing proprietary access. It uses external sensors to track hands, feet, and video, validated on systems like the da Vinci Xi. The system achieved gesture recognition performance comparable to internal robot telemetry and is released with annotated datasets, including the first for hernia repair suturing on high-fidelity models.