Researchers at Johns
Hopkins University have successfully trained a robot to perform
surgeries at a skill level comparable to human surgeons. Using imitation
learning, they taught robots to execute complex surgical procedures
without human intervention. Remarkably, doctors can interact with these robots
just as they would with a real surgical resident.
A Milestone in Robotic Surgery
The robot has acquired its
skills by observing experienced doctors perform surgeries. This development is
significant because, in the future, such robots could dramatically increase the
number of surgeries performed in less time, addressing healthcare bottlenecks.
Demonstrating Advanced Capabilities
At a recent conference in
Munich focused on robotic learning—an influential event in robotics and machine
learning—researchers showcased their work with the da Vinci Surgical
System robot. They used imitation learning to train the robot on three
fundamental tasks:
- Manipulating needles.
- Handling body tissues.
- Performing sutures.
This training approach
eliminates the need to program every robotic movement manually, pushing robotic
surgery closer to full autonomy.
"A Magical Model"
Assistant Professor and
senior author Axel Krieger from Johns Hopkins University's
Department of Mechanical Engineering described the model as
"magical." He explained, "We simply provide it with camera
input, and it predicts the robotic movements needed for surgery. The model is
so capable that it learns tasks we haven't explicitly taught it. For example,
if it drops a needle, it autonomously picks it up and continues the
procedure."
Enhanced Automation with Human Interaction
Unlike conventional
robotic-assisted surgeries, where doctors control robots with joystick-like
controllers, these advanced robots can now interpret and correct their
movements independently. This new capability allows surgeons to interact with
robots using natural language commands like, "Move left," "Move
right," or "Perform this task," akin to guiding a human trainee.
Training Inspired by AI Models Like ChatGPT
The training process for
these surgical robots parallels the development of large language models like
ChatGPT. Instead of learning from words and images, the robot was trained using
videos recorded by wrist cameras attached to the da Vinci Surgical System
during surgeries.
Reducing Time-Consuming Programming
Previously, programming
surgical robots required meticulous hand-coding for each movement, a
time-intensive process. With this breakthrough, surgical robots can learn
directly from video demonstrations, significantly streamlining their
development process.
This innovation marks a
pivotal step toward autonomous robotic surgery, promising a future where
advanced surgical robots can enhance medical outcomes and accessibility
worldwide.