In this study, we present a control approach based on reinforcement learning applied to a magnetically flexible endoscope (MFE).
This device, designed to reduce pain and increase ergonomics in colonoscopy, is composed of one external permanent magnet (EPM) and one internal permanent magnet (IPM) in the tip of the endoscope.
The aim of this work is to guarantee an autonomous waypoint tracking of the endoscope able to navigate the entire colon, simultaneously maintaining contact between the endoscope and the tissue.