Robot Masters Surgical Tasks Simply by Watching Videos

Robot Masters Surgical Tasks Simply by Watching Videos

Performing surgery takes years of training for humans, but a robot could learn the skill more easily with today’s AI technology. Researchers from Johns Hopkins University (JHU) and Stanford University have taught a robot to perform various surgical tasks just by watching videos of the procedures.
With the help of an AI model trained on videos of surgery videos, a robot system has successfully carried out difficult surgical tasks as skillfully as a human
Intuitive

Performing surgery takes years of training for humans, but a robot could learn the skill more easily with today’s AI technology. Researchers from Johns Hopkins University (JHU) and Stanford University have taught a robot to perform various surgical tasks just by watching videos of the procedures.

The team used a da Vinci Surgical System, a robot typically controlled by a surgeon, which allows for precise movements like dissection and suturing. The system costs over $2 million, not including accessories or training.

The da Vinci surgical system at work
Intuitive

Using imitation learning, the researchers trained the robot to perform tasks such as manipulating a needle, lifting tissue, and suturing. Remarkably, the robot could perform these tasks as well as humans and even correct its own mistakes, like picking up a dropped needle automatically.

The AI model combines imitation learning with the architecture used in chatbots like ChatGPT. Instead of processing text, it outputs kinematics, a mathematical language that directs the robot’s arms. The model was trained on hundreds of videos filmed from wrist cameras on da Vinci robots during surgeries.

Surgical Robot Transformer Demo

New AI Method and robots Could Accelerate Autonomous Surgery, Reducing Errors and Improving Accuracy

The researchers believe their method could rapidly teach robots to perform any surgery, making the process far easier than traditional hand-coding each step. According to JHU’s Axel Krieger, this approach could accelerate the path to autonomous surgery, reduce errors, and improve accuracy.

This innovation could be a major breakthrough in robot-assisted surgery. While some robotic systems, like Corindus’s CorPath, already assist with certain surgical steps, they lack full autonomy. Krieger noted that traditional coding for robotic tasks is slow, often taking years to model even one action.

‘Wrist’ cameras attached to the arms of the robot surgical system capture footage to help train the AI model
Johns Hopkins University / Stanford University

In 2022, Krieger’s team developed the Smart Tissue Autonomous Robot (STAR) at JHU, which performed suturing without human assistance. Now, the JHU team is working on teaching robots to perform complete surgeries, though it will likely be years before robots fully replace human surgeons. However, advancements like this one could make surgeries safer and more accessible worldwide.


Read Original Article: New Atlas

Read More: Scitke

Share this post

Leave a Reply