Robotic Assistants in the Operating Room: A Systematic Review of Vision‐Based Systems for Nonoperative Tasks

Nonoperative tasks in the operating room, such as instrument handling, sterile packaging, and logistics, are demanding yet have received less robotics attention than autonomous surgery. This review, the first to synthesize this literature, maps the perception methods, autonomy levels, methodological quality, and clinical readiness of vision‐based robotic assistants for scrub nurse, circulating nurse, and OR logistics roles. Searches of PubMed, IEEE Xplore, Scopus, ACM Digital Library, Web of Science, and Google Scholar returned 465 records; 20 publications describing 11 robotic system families met inclusion criteria. Most evidence, 17 publications across eight system families, concerns scrub nurse assistance; circulating nurse support, OR logistics, and dental assistance are each represented by one system. Vision methods have moved from classical image processing and predefined gestures toward deep learning, multimodal perception, and vision–language–action models, but autonomy has not kept pace: one system family reaches LoA 1, seven LoA 2, and three LoA 3, with Technology Readiness Levels spanning TRL 4 to TRL 6. Reported success rates are often high but rarely comparable and mostly generated in laboratory or simulated settings. Closing that gap requires stronger evidence on safety, sterility, failure recovery, workflow integration, execution speed, reproducibility, and, notably absent, manipulation of deformable sterile materials.