AI-trained robots that imitate skilled surgeons

By doing away with the need to manually configure robots for every single movement during a medical procedure, this innovation brings robotic surgery one step closer to complete autonomy. Complex procedures may eventually be carried out autonomously by robots.

The da Vinci Surgical System robot was taught three basic activities by the research team using imitation learning: stitching, lifting tissue, and using a needle. Hundreds of films of surgeries conducted by human surgeons were fed into the model, which taught the robot to perfectly reproduce these procedures.

This technology greatly speeds up the training process by allowing the robot to learn from demonstration movies, in contrast to traditional programming methods that require hand-coding each step of a procedure.

The main novelty is merging machine learning architecture akin to ChatGPT with imitation learning. However, the model uses kinematic data to convert the robot’s actions into mathematical computations rather than text.

As a result, the robot can anticipate the required motions based on camera input without requiring specific instructions for each action.

The da Vinci system has been criticized for its imprecision despite its widespread use. By teaching the robot to make relative movements rather than absolute ones, the researchers were able to compensate for the system’s errors. Consequently, the robot might be able to apply what it has learned to different settings and circumstances, self-correcting errors like spilling a needle.

With this innovation, surgical robots might learn a variety of techniques in a matter of days, significantly cutting down on the amount of time needed for training. The group is now attempting to expand this method to allow robots to carry out full surgeries on their own, which could improve surgical precision and lower medical errors.