Researchers at Johns Hopkins University have trained a robot to perform intricate surgical tasks at par with human surgeons by observing experienced doctors in action. At the Conference on Robot Learning in Munich, the team revealed their use of imitation learning to train the da Vinci Surgical System robot in three essential tasks: needle manipulation, tissue lifting, and suturing. This approach eliminates the need to manually program each robot movement, advancing robotic surgery toward greater autonomy.

Axel Krieger, an assistant professor in JHU’s Department of Mechanical Engineering, described the model’s capabilities as "magical," highlighting its ability to predict necessary robotic movements based solely on camera inputs. Remarkably, the robot can handle unexpected scenarios, such as dropping a needle and picking it up independently—a skill not explicitly taught.

Unlike traditional robot-assisted surgery, where doctors guide robots using joystick-like controls, this new system allows robots to comprehend and rectify errors autonomously. The training method resembles the development of large language models like ChatGPT but uses video data captured from wrist cameras on the da Vinci robots during procedures.

Previously, programming surgical robots involved labor-intensive coding for every individual action. However, the new system simplifies this process, enabling doctors to communicate with the robot conversationally, issuing commands like “Move left” or “Perform this task.”

While these advancements are groundbreaking, experts caution that the high stakes of surgery mean fully autonomous robots may still require years of refinement before widespread adoption.