In July 2026, Nature published a breakthrough: a UC San Diego team had doctors teleoperate a Unitree G1 humanoid to perform two laparoscopic gallbladder removals on live pigs — the first time a general-purpose humanoid did the main steps of standard laparoscopic surgery on a living subject.
The trigger was Elon Musk. After his April 2025 X post claiming robots would beat top surgeons within five years, the team decided to test it. Lead author Liang Zekai says the post was the direct catalyst.
The setup was deliberately generic. They added only a hand interface and used standard wristed laparoscopic instruments built for human surgeons — no custom robot. The G1 costs under $20,000 and is smaller than an adult; da Vinci systems run past $1 million. The contrast opened the imagination: a humanoid in a remote clinic, desert or battlefield.
But it was a feasibility test. The robot could not operate autonomously; a doctor teleoperated throughout, with repeated pauses to recalibrate and a preventive harness strapped on. Surgeons scored the system’s clinical maturity a median 2.5 out of 5.
The limits are physical. The G1 stands about 1.3 metres with roughly 450 millimetres of arm reach versus 1.6 to 1.8 metres for an adult; the second procedure saw minor bile spillage and liver-bed bleeding, handled by suction and cautery. Bench error was about 1.3 millimetres straight, 10.4 millimetres on complex curves. Sterility meant putting the robot in gloves — the industry must set standards first.
Liang rejects Musk’s five-year claim: true autonomous surgery needs anatomy recognition, contingency handling and liability rules. The realistic near-term role is hospital support — transport, scope-holding, retraction, equipment ops. The experiment did not birth a robot surgeon. It moved surgery into the job description of a general-purpose humanoid.
Original report: OFweek Robotics.
Translated and adapted from OFweek Robotics (robot.ofweek.com).