In a significant advance for robotic surgery, researchers at Johns Hopkins University have introduced the SRT H (Surgical Robot Transformer–Hierarchy) system, which successfully performed eight full laparoscopic gallbladder removals on pig organs in a laboratory setting. Reported on July 9, 2025, this marks the first demonstration of step-level autonomy for a multi-stage soft tissue procedure, setting the groundwork for future in vivo and human applications.
It’s important to clarify while often described as “autonomous,” the procedures occurred on ex vivo (cadaveric or isolated) pig organs and involved passive human supervision, including manual tasks like instrument reloading. Still, the AI made surgical decisions, adjusted for visual variation, and completed the procedure independently once activated, making this a remarkable stride toward surgical autonomy.
How the SRT H System Works
The SRT H uses a hierarchical AI model inspired by large language models, including transformer architectures similar to those used in tools like ChatGPT. However, this system specializes in surgery and includes:
• Imitation learning: Trained on over 17 hours of annotated surgical video to understand procedural intent.
• Language-conditioned planning: Converts instructions like “clip the cystic duct” into real-world tool maneuvers.
• Visual feedback loop: Reacts to visual anomalies such as tissue movement or obstructions.
• Voice override interface: Allows surgeons to provide verbal corrections or commands during operation.
In all eight tests, SRT H autonomously handled key tasks such as identifying anatomical structures, positioning surgical clips, and completing gallbladder removal—while making 6–10 real-time corrections per procedure. These represent self-directed micro-adjustments, not full procedural reboots, underscoring the model’s resilience.
Why This Matters
1. Realistic Simulation of Soft-Tissue Autonomy
This is the first system to navigate a complex soft tissue procedure in full, on realistic anatomical models, with autonomous correction and dynamic adaptation—previous robots relied on rigid scripts or were restricted to simpler tasks.
2. Valid Platform for Pre-Clinical Progression
Though not performed in live animals, the results strongly suggest readiness for controlled in vivo trials, expected to begin later in 2025 or 2026. Live trials will test the robot’s ability to adapt to bleeding, respiration, and organ motion—factors absent in cadaveric tissue.
3. AI-Enabled Surgical Insight, Not Replacement
SRT H’s transformer-based planning logic allows it to generalize across variable anatomy, a necessary trait for future human applications. But it’s not designed to replace human surgeons—it functions as a scalable assistive platform, capable of standardizing basic procedures and increasing access.
• Surgeons & OR Teams: Shift from technical operators to procedure supervisors and AI governance experts.
• Hospitals & Health Systems: May see gains in throughput, consistency, and training cost reductions.
• Patients: Could benefit from consistent outcomes and increased procedural access in low-resource regions.
• Medical Schools: Must integrate new curricula around AI supervision, robotic collaboration, and data-informed care delivery.
Dr. Axel Krieger, project lead at Johns Hopkins, stated:
“We’ve moved beyond isolated robotic gestures to full, adaptable procedure logic. This project isn’t about removing surgeons—it's about supporting them and expanding access to safe operations.”
Dr. Julie Wu, a practicing minimally invasive surgeon, emphasized:
“It’s impressive that the robot can recognize when it’s slightly off and correct without external input. But we need to see how it behaves with real bleeding and tissue resistance—those are make-or-break factors in live surgery.”
What’s Next?
Milestone Timeline
Live-Animal Trials 2025–2026
Human Pilot Studies (Supervised) Within 5–10 years
Regulatory Pathway (FDA) 2030 and beyond
Procedure Expansion Hernia, colectomy, gynecologic surgery
To prepare for this next chapter in surgical innovation:
• Hospitals: Begin building AI-ready infrastructure—connected imaging systems, real-time analytics, and simulation platforms.
• Educators: Train surgical teams in robotic oversight, procedural auditing, and AI risk management.
• Tech Developers: Follow ongoing live trials and collaborate to build safe, interoperable systems.
• Regulators & Ethics Boards: Develop early frameworks for human–robot accountability, surgical audit trails, and clinical risk standards.
Autonomy in surgery is no longer conceptual, it’s becoming practical. While SRT H hasn’t yet operated on a live subject, it’s arguably the closest any soft-tissue system has come to a clinically relevant, fully autonomous procedure.