Yorg Healer
December 11, 2024 08:45
Researchers at Johns Hopkins University and Stanford University have developed robotic surgery by integrating AI, allowing robots to perform complex surgical tasks autonomously and with precision.
In a significant advancement in medical technology, researchers at Johns Hopkins University and Stanford University have improved the capabilities of robotic surgery, allowing robots to perform complex surgical tasks autonomously. These developments, reported by NVIDIA, signal a potential change in the way surgery is performed globally.
AI integration into robotic surgery
The team integrated a visual language model (VLM) trained on extensive surgical video footage with the da Vinci robotic surgery system. This integration allows the system’s robotic ‘hands’ to autonomously perform critical surgical tasks, including tissue manipulation, needle handling, and suturing.
Traditionally, robotic systems required detailed programming for each movement. However, the new model leverages imitation learning to allow the robot to replicate movements observed in surgical videos. This approach represents a paradigm shift in robotics, as noted by Jiwoong “Brian” Kim, a postdoctoral researcher at Johns Hopkins University who highlighted the imitative learning potential of autonomous surgical robots.
Technical achievements and experiments
Researchers trained the model utilizing NVIDIA GeForce RTX 4090 GPUs, PyTorch, and CUDA-X libraries. The findings were presented at the Robotics Learning Conference in Munich and highlighted the capabilities of the da Vinci Surgical System, which is widely used in laparoscopic surgery worldwide.
For training, a small camera was attached to the robotic arm to capture over 20 hours of surgical procedures. This data contains accurate kinematic information that is important for VLM training. The experiments were conducted using animal flesh, and the robot performed almost flawlessly in a zero-shot environment, solving unexpected problems autonomously.
Future Implications and Developments
The success of these experiments points to a future in which autonomous robotic surgery may become commonplace. Kim and his team are already conducting additional experiments using animal cadavers and expanding their training data to improve the capabilities of the robotic system.
These advances are likely to impact the future of surgery, potentially improving precision and reducing the risk of human error. To learn more about this groundbreaking research, visit the NVIDIA blog.
Image source: Shutterstock