Overview

• SRT-H achieved flawless ex vivo gallbladder removals on pig cadavers under eight different anatomical and simulated emergency conditions, as reported in Science Robotics.
• The system mastered the 17-step procedure by studying 34 pig cadaver surgeries recorded over 17 hours and receiving targeted language-conditioned feedback.
• Unlike its STAR predecessor, SRT-H requires no tissue markers or fixed surgical plans and self-corrects its actions without human intervention.
• Researchers noted that the robot’s real-time adaptability allows it to identify critical structures, place clips and cut tissue while responding to unexpected variances.
• The Johns Hopkins team is now applying the same imitation-learning framework to train SRT-H on additional procedures ahead of planned human clinical trials.