Robotically assisted surgery is increasingly popular in many healthcare institutions around the world, providing valuable benefits such as shorter operating times, reduced radiation exposure for patients and practitioners and increased operational accuracy. It has also allowed previously more invasive surgeries to be performed laparoscopically, reducing the risk of infection and trauma to the patient thereby shortening hospital stays. The advantages achieved through progress in robotic surgery speak for themselves.
Various robotic systems are being used in robot-assisted procedures1. For example, in January 2026, Intuitive secured FDA clearance for the da Vinci (DV5) to be used for several cardiac procedures. Meanwhile similar systems such as MedBot’s Toumai has gained FDA-IDE clearance for use in clinical trials and research.2
July 2025 to January 2026 – Developments and trials in autonomous robotic surgery
Despite these advancements, fully autonomous robotic surgery without any human intervention seems a distant goal. Or it did until July 2025, when a research team at John Hopkins University conducted an experiment, successfully removing unseen gallbladders from eight ex vivo pigs using an autonomous robotic system without any human intervention3. The system was able to repeatedly correct mistakes and autonomously respond to anatomical variation. A Hierarchical Surgical Robot Transformer is trained using images and language annotations, avoiding reliance on depth sensors and specialised hardware. It further includes a corrective language interface to enable human surgeons to intervene and correct any behaviours in a timely manner during a procedure4. This research presents a great stride in autonomous robotic surgery and safety mechanisms for autonomous robotic surgical systems.
This is a milestone in the move towards autonomous surgery and safer and more efficient healthcare aided by AI and robotics. Some healthcare professionals and researchers have indicated that trials for fully autonomous robotic surgery on humans look like a serious possibility in the next decade5.
Although no peer reviewed papers have been published on the trial as of yet, there have been reports that in December 2025, a team at Shanghai MicroPort MedBot successfully performed an abdominal surgery on an in vivo pig using the Toumai – a world-first for in vivo robotic surgery without human intervention. In this experiment, the company’s Toumai surgical robot, powered by a proprietary multimodal AI model called Neuron, carried out the removal of the gallbladder and associated procedures. The AI model was trained on 23,000 recorded surgeries, enabling it to simulate surgical decision-making and adapt to the patient’s anatomy in real time. The robot autonomously completed 88% of the surgical steps on the first attempt and adjusted itself to finish the remaining steps with minimal correction6. This achievement marks a significant step towards fully autonomous surgery on live patients. It provides a proof of concept for overcoming complex issues such as blood flow and patient movement during surgery. Shanghai MedBot emphasised that while this pig trial was a milestone for surgical AI, the system is not yet approved for human use and will require further validation and safety testing7. Nonetheless, the result showcases how cutting-edge large-model AI can drive surgical robots to operate with a high level of independence, handling required corrections and variability much like a trained surgeon.
A recently as January 2026, a team at the Institute of Automation of the Chinese Academy of Sciences published their results of another significant breakthrough in the field of autonomous surgical robotics. They developed the Autonomous Robotic System for Intraocular Surgery (ARISE), which addresses the longstanding challenges of restricted visual perception and depth estimation in microscale eye procedures. ARISE leverages two key innovations to achieve intelligent perception and high-precision instrument control within the intraocular space: a multiview spatial fusion algorithm that reconciles imaging disparities and corrects dynamic spatial misalignments, and a criterion-weighted fusion of multisensor data that mitigates inconsistencies in detection range, error magnitude, and sampling frequency. In a series of trials, ARISE achieved 100% success rates in subretinal, central retinal vein, and branch retinal vein injections across ex vivo porcine and in vivo animal models. Compared to manual and teleoperated approaches, ARISE reduced positioning errors by 79.87% and 54.61%, respectively, demonstrating its potential to enhance surgical consistency, reduce training time, and enable safe, precise interventions in both clinical and remote settings8.
Conclusion
Challenges certainly remain before we see robots operating on humans without oversight; however, researchers say that such trials are no longer science fiction but a credible goal for the next decade. Each success – from Johns Hopkins’ autonomous robotic gallbladder removals to The Institute of Automations autonomous robotic retinal injections – adds momentum.
These developments highlight a rapidly accelerating race toward fully autonomous surgery. In just the past year, the field has moved from speculative to a proof-of-concept in excised pig organs to autonomous operations in living animals. The rapid progress demonstrates significant government and industry backing. Advancements in a range of technical fields are driving this progress, for example, robotics hardware, AI integration, multimodal prompting and processing, computer vision, and training data collection and generation. In such an active and fast-paced environment it is essential to consider IP strategy early in development. The number of patent filings for AI assisted MedTech solutions is increasing quickly, as is further analysed and discussed in our report on how AI is shaping the future of MedTech. We look forward to seeing how these technologies progress as AI continues to develop and medical research finds new ways to make use of it.
References
-
https://www.massdevice.com/intuitive-fda-510k-clearances-cardiac-procedures-da-vinci-5/
-
https://hub.jhu.edu/2025/07/09/robot-performs-first-realistic-surgery-without-human-help/
-
https://www.techinasia.com/news/chinas-medbot-tests-autonomous-ai-surgery-robot-on-pig
-
https://www.science.org/doi/abs/10.1126/scirobotics.adx7359?af=R
Kara is a patent technical assistant in the Engineering practice group. Kara specialises in Software, Machine learning, Mobile robotics and Path planning algorithms.
Email: kara.quast@mewburn.com
