Surgical Robotics & AI: From da Vinci 5 to Autonomous Surgery

The da Vinci system, the icon of robotic surgery, has transformed minimally invasive surgery since the early 2000s. The underlying logic, however, did not change: the robot was a tool that transmitted the surgeon's hand movements under a magnified view, with tremor filtering and increased range of motion — that is, teleoperation. The deciding, cutting, and suturing were all done by a human. AI now enters this equation from three angles: smarter devices, robots that learn from video, and supervised autonomy.

Smarter Hardware: da Vinci 5

Intuitive Surgical's next-generation platform, da Vinci 5, received FDA clearance and entered use at numerous U.S. centers in 2024–2025. The system has more than 150 design innovations and 10,000 times more compute power than its predecessor. The signature new feature, presented as "first of its kind," is force feedback: the surgeon can now feel the tension being applied to tissue.

da Vinci 5 also offers advanced data analytics and AI-assisted feedback features that, for now, provide decision support and workflow improvements to the surgeon rather than taking over the operation. In other words, da Vinci 5 is still fully under the surgeon's control — but increasingly aware of context.

Robots That Learn by Watching Video

In 2025, the concept of "physical AI" came to the fore in surgery via multiple approaches: da Vinci robots learning from surgical videos, systems such as Moon Surgical providing collaborative tableside assistance, and STAR's supervised autonomy. The underlying paradigm shift is imitation learning: rather than programming the robot for every possibility one by one, skills are acquired by watching expert surgeons.

Vision in the OR: Image-Guided AI

Before autonomy comes a much more widespread contribution: enhancing the surgeon's vision during the procedure. Computer-vision algorithms can analyze the live image from a laparoscopic or robotic camera in real time and highlight anatomical structures (vessels, nerves, organ boundaries). This is a valuable safety layer, especially for operations requiring a "critical view of safety" — to prevent serious complications such as cutting the wrong duct.

Image-guided approaches align preoperative CT/MRI data with the live intraoperative view (registration), offering the surgeon a kind of "road map." AI both automates that alignment and tries to keep the map up to date as tissues shift. These capabilities mostly remain at the decision-support level — they do not steer the robot but enrich what the surgeon sees — yet they are precisely the visual-understanding foundation that opens the door to autonomous steps.

The Frontier of Autonomy: STAR and Johns Hopkins SRT-H

The pioneer of autonomous tissue manipulation has been STAR (Smart Tissue Autonomous Robot). In 2020, STAR performed a laparoscopic bowel suturing procedure in a living animal largely autonomously for the first time; with computer vision and machine learning, it could adapt to tissue deformation.

The real leap came in 2025. In a study by a Johns Hopkins team published in Science Robotics in July 2025, a system called SRT-H performed the complex stage of a gallbladder removal on a realistic phantom without human intervention. The task comprised 17 sequential subtasks requiring identification of specific ducts and arteries, precise grasping, strategic placement of clips, and cutting with scissors — over the course of minutes. The system reported 100% success across eight different (previously unseen) porcine gallbladders.

SRT-H's innovation is that it learned the gallbladder task by watching videos of Johns Hopkins surgeons performing the procedure on porcine cadavers. Visual training was reinforced by captions describing the tasks. Unlike previous rigid, preprogrammed systems, SRT-H learns by observing experts and can correct its own errors in real time.

Levels of Autonomy: Where Do We Stand?

Surgical autonomy is helpful to think about in graded levels, much like self-driving cars. On one end is fully surgeon-controlled teleoperation (da Vinci); on the other, full autonomy; intermediate steps such as "supervised autonomy" lie in between. Today, the overwhelming majority of FDA-cleared surgical robots remain at low levels of autonomy — that is, under human control.

Tempering the Excitement: Important Limits

SRT-H's success is historic, but the context must be set correctly. The study was performed on porcine cadavers and on one stage of cholecystectomy. In living tissue (in vivo) and ultimately in humans, anatomy that varies from patient to patient, bleeding, and inflammation will pose much greater challenges. So the "robot performed surgery on its own" headlines reflect one stage in a controlled research environment; routine clinical autonomous surgery remains far off.

Moreover, as autonomy increases, questions of responsibility, error management, regulatory clearance, and patient safety become even more critical. Today's reality is this: AI is not yet replacing the surgeon; it is giving the surgeon a sense of touch, sharper vision, and — for select steps — a smart assistant that can be entrusted with parts of the procedure. Surgical robotics is moving from "the human's tool" toward "the human's supervised partner."