AI has made our lives easier by unburdening employees from doing a routine task and allowing them to concentrate on more complex work. But it is not just limited to doing paperwork, or handling redundant data.
Scientists taught a robot how to do sutures. They did so by feeding the robot 78 slowed down films related to medical procedures. The research included connecting a pair of automated pincers to the AI algorithm to use a needle and thread.
The aim behind “imitation learning” was to train AI in a way similar to semi-autonomous cars.
After analyzing the video and learning the motion, the researchers broke down the suturing task into programmable actions:
- Needle insertion
- Needle extraction
- Needle hand-off
Researchers claimed that this enabled Motion2Vec, their deep learning system, to guide the robot hands to stitch a fake wound with almost 85% accuracy.
This breakthrough could mean surgeons could focus on more complex procedures, while robots could perform these repetitive tasks under a surgeon’s supervision.
The current stitching system has a 9.4mm error rate; hence it is not ready for real patients.
#AIMonks #AI #Robots #Surgeons #Automation #Suture #ImitationLearning
0 Comments