Two Unitree G1 robots perform live surgery in a new study by the University of California, San Diego.
ARClab
In a global first, two humanoid robots have performed surgery on a live patient. The robots performed laparoscopic cholecystectomies, or gallbladder removals: retracting tissue, dissecting, clipping and lifting the gallbladder out of the liver bed. The two big caveats: the patient was a pig and the robots were fully teleoperated by trained human surgeons. Another caveat: as you can see in the image above, the robots were tethered so they could not accidentally fall and injure their porcine patient.
Still, it’s an impressive accomplishment and an important milestone.
This was essentially an experiment performed by a team from the University of California, San Diego. The goal: understand whether humanoid robots are useful for extending medical care in the physical absence of human doctors. The answer, ultimately, is not yet … but probably soon.
“Humanoid form factors offer unique potential, particularly for assisting with surgical tasks,” the team reported in its Github overview. “Traditionally, robotic systems for surgery are purpose-built platforms such as Intuitive Surgical’s da Vinci Surgical System, and it remains unclear how close current humanoid systems are to meeting the precision, control, and safety requirements of minimally invasive surgery.”
What they learned is genuinely significant: a humanoid robot completed a real surgical procedure on living tissue. That said, humans were still deeply involved, and the team’s paper in Nature states that there are still “key technical challenges that must be addressed before clinical deployment.”
The team from UC San Diego’s school of medicine and ARClab, an advanced robotics lab at the university, did not custom-build a robot for this job. Rather, they used two off-the-shelf Unitree G1s. These are small, cheap robots: four to five feet tall, only about 70 pounds and purchasable for under $20,000, although it’s likely these specific robots had Unitree’s three-fingered “Dex3” dexterous hands that come with a thumb, index and middle finger, each with multiple actuated joints. They also have extra wrist and waist articulation, making this version capable of holding and manipulating tools. It also ships with 3D LiDAR and a depth camera: very useful for surgery tasks.
This in itself is cause for optimism, if you are interested in having humanoid robots perform surgery in the near future: they used a super-cheap, basic robot. If they used a more advanced robot – perhaps 1X’s Neo, with 25 degrees of actuated freedom hands – the surgery would be even more doable.
That said, the benchmark here is Intuitive Surgical’s da Vinci, the robotic system that has defined the category for two decades and sits in thousands of operating rooms. But da Vinci is a fixed, purpose-built platform. It is wheeled in, docked, and it does one thing extremely well. It does not walk down a hall and it’s probably not affordable for a small city or county hospital, because it costs on the order of a couple million dollars.
The humanoid bet is different.
It’s unlikely that a general-purpose robot would beat da Vinci at surgery, but it could, in principle, move through a hospital built for humans, use tools designed for human hands, and take on many embodied tasks rather than one. Plus, roughly half the world lives in a place with too few surgeons, and “medical desert” isn’t only a developing-world problem: rural regions across the U.S. and Canada routinely send patients hours away for procedures a robot and a remote specialist could handle locally.
Plus there are other edge cases: a winter-over crew in Antarctica … an astronaut on the International Space Station … a voyager on a future Mars transit mission.
A low-cost, general-purpose humanoid that can be teleoperated by an expert thousands of miles away is not just a hospital gadget. It could be essential infrastructure for places medicine can’t currently reach.
Ultimately, we’ve learned that even basic, cheap humanoid robots can perform surgery. Today it’s teleoperated, sure. Tomorrow … who knows? Maybe we’ll have robots that can download a surgical physical AI foundation model for a small fee and – Shazam! – you’ve got a surgeon.
With health care costs and availability where they are right now, surely that’s a potential positive at least.