The team from Johns Hopkins and Stanford Universities has developed an artificial intelligence-based robot capable of learning surgical gestures through imitation.
Unlike traditional methods that require exhaustive programming, this surgical robot acquires its skills by watching videos of real surgeries. This breakthrough was presented at the latest
Conference on Robot Learning.
The model leverages a technology inspired by the architecture of artificial intelligences like ChatGPT. Here, no words or phrases: it decodes movements and translates them into precise robotic actions. A true leap forward in the realm of medical robotics.
To conduct this work, the researchers used the Da Vinci surgical system, already widely deployed in hospitals. Equipped with miniaturized articulated arms, this robot excels in minimally invasive procedures. However, its traditional programming remains tedious and limited. The new approach promises to overcome these constraints.
Hundreds of videos from cameras installed on Da Vinci robots were used as learning material. These recordings, often used for post-operative analysis, became training tools for the robot. As a result, it can now master basic tasks such as handling needles or suturing, with dexterity approaching that of human surgeons.
A remarkable feature of this system is its adaptability. For instance, if a needle falls, the robot retrieves it without human intervention. This ability goes beyond what it was explicitly taught, highlighting the potential of this technology.
The team is already looking further ahead: training the robot to perform a complete surgical procedure. According to the researchers, it would only take a few days to teach it a complex procedure, thus accelerating robotic autonomy in the medical field.
With approximately 7,000 Da Vinci robots operating worldwide, this imitation learning method could transform modern surgery. By democratizing expert gestures, it paves the way for more precise and accessible procedures, while reducing the training time for medical teams.
In the near future, these robots could expand into other surgical specialties, bringing unprecedented precision and safety to medicine.
Article author: Cédric DEPOND