Two operations inside a surgical training center at the University of California San Diego ended the same way: a pig's gallbladder out, and a five-foot machine standing where a bedside assistant usually stands. The robot weighed 60 pounds. It held a laparoscopic instrument in each hand. It did not decide to do any of it.
That last detail is the one most likely to get lost. The machines were teleoperated, meaning a human surgeon sat at a console and moved their arms in real time. Engineers and surgeons at UC San Diego published the results in Nature, and Nature Medicine flagged the work in a research highlight three weeks later. Both procedures were performed on female pigs under general anesthesia, and neither had to be converted to conventional laparoscopy or open surgery.
Sixty Pounds of Robot Next to an 1,800-Pound Alternative
The comparison that makes this interesting is not robot-versus-human. It is robot against the robots hospitals already own.
A conventional surgical system carries three or four purpose-built arms, runs proprietary instruments and software, and weighs about 1,800 pounds. Operating rooms usually have to be retrofitted around one. The machines used here, nicknamed Surgie, are five feet tall and mobile, and the team built adapters so they could grip ordinary laparoscopic tools instead of custom ones. The surgeon drove them from a console using hand controllers, a stereo headset, and a foot pedal.
"We were surprised at how well Surgie meshed with our workspace and workflow," said Nikita Thareja, a general surgery resident and a co-author, in the university's account.
Ryan Broderick, interim director of the university's Center for the Future of Surgery, told ABC News that the footprint was the surprise. "The space constraints didn't exist like in traditional robotic surgery," he said. "It was a human-type bedside assistant, so it just fit into the space that we're traditionally used to being in for laparoscopic surgery."
The Word Missing From This Milestone Is Autonomous
In one operation, a teleoperated humanoid handled the instruments while a human surgeon assisted at the bedside. In the other, two humanoids worked side by side, one running the instruments and the other helping with the camera and tissue retraction. That paired arrangement was used only briefly, and human assistants handled most of the bedside work in both cases.
Both procedures followed a standard cholecystectomy sequence: retraction, dissection, confirmation of the critical view of safety, clipping the cystic duct, and separating the gallbladder from the liver bed. At no point was the robot making its own decisions. Autonomy is the stated destination, not the achievement.
"One of our goals is to develop the autonomous surgical assistant," said Michael Yip, a UC San Diego electrical and computer engineering faculty member and one of the paper's senior authors. "Many communities struggle with adequate staffing on the surgical team, which means patients are not being treated."
Work on genuinely autonomous surgery is happening elsewhere, on a separate track. A Johns Hopkins team led by Axel Krieger, with lead author Ji Woong Kim, trained a system on surgical video and had it complete the clipping and cutting phase of a gallbladder removal on pig cadavers with nobody at the controls; the work was published in Science Robotics. That machine was not humanoid. The San Diego robots are humanoid but not autonomous. Nobody has combined the two in a living animal.
Why the Operations Took Longer Than They Should Have
The paper is unusually candid about what went wrong, which is the most useful part of it for anyone judging how close this is to a hospital.
The robots had to be recalibrated and repositioned repeatedly, causing interruptions that lasted more than three minutes while the team realigned the arms with the surgical ports. Both operations took considerably longer than the same surgery on an established platform. Roughly 156 milliseconds passed between the operator's hand moving and the robot responding, above the sub-150-millisecond delay the field generally considers desirable. Straight tool movements were accurate to about 1.3 millimeters, but curved motions drifted by more than 10. Surgeons also reported intermittent overheating, limited arm strength, and restricted range of motion.
The first operation finished without major complications. During the second, minor bile leakage and bleeding from the liver bed occurred, and the team controlled both with suction and electrocautery. Neither case met the study's predefined criteria for converting to another approach.
Shanglei Liu, an assistant professor of surgery who teleoperated during the study, framed the slowness as a stage rather than a verdict. The first robotic laparoscopic surgery took six hours, he noted. It now takes about 30 minutes.
Ships, Villages and the Access Problem Behind the Project
The argument for putting a walking robot in an operating room is not that it cuts better. It is that it can go places a 1,800-pound console cannot.
"It's a fraction of the cost, and it takes a fraction of the space in an operating room," Liu said. "So it's easy to deploy, anywhere from rural areas to the battlefield, and even to space." Yip has pointed to remote communities with thin surgical staffing and to disaster response, where field medicine has to be stood up quickly.
Readers should keep the evidence in proportion. This was a feasibility study in two animals, run at a surgical simulation center with full laparoscopic infrastructure already in place and an experienced surgeon at the console. Sterility remains unsolved: the team covered the robot arms with gloves, which does not meet the standard required for human surgery, and current commercial humanoids lack components that can be autoclaved. No human has been operated on by any of these machines, and no regulator has cleared any of them. Surgical robotics also has a formal evaluation pathway, the IDEAL framework, at the earliest stage of which this work sits.
The work was funded in part by the National Science Foundation and the National Institutes of Health. Two of the authors are surgeons, and Yip co-founded a robotics company.
What the study establishes is narrower and still notable: general-purpose humanoid hardware, bought largely off the shelf, can hold ordinary surgical tools and complete a real operation inside a living body when a trained surgeon is steering.
Key Questions Answered
What actually happened? Surgeons at UC San Diego used teleoperated humanoid robots to complete two laparoscopic gallbladder removals in live pigs. One surgery paired a robot with a human assistant; the other briefly used two robots together.
Were the robots operating on their own? No. A human surgeon controlled every movement from a console. Developing an autonomous surgical assistant is a stated goal of the team, not something demonstrated here.
Has this been done in people? No. This was a preclinical animal study involving two procedures. There is no approved human use of these machines for surgery.
Why use a humanoid robot instead of existing surgical robots? Existing systems weigh about 1,800 pounds and often require modifications to operating rooms. The humanoids used here are five feet tall, weigh 60 pounds, are mobile, and can be adapted to hold standard laparoscopic instruments.
What went wrong during the procedures? The robots needed repeated recalibration, control lagged by about 156 milliseconds, curved movements were imprecise, and the arms overheated and lacked strength. The second operation involved minor bile leakage and liver bed bleeding, both controlled.
What would need to change before patients see this? Latency, calibration stability, precision and sterilization all need substantial work, followed by regulatory review and human trials. Nothing about this study changes current surgical care.
