Abstract
This study aims at developing an autonomous risk detection system to improve the safety of robotic surgery. Conventional robotic laparoscopic surgery is conducted using a master-slave robotic system, which relies on communication stability between the master and slave sites. Communication delay or errors may lead to unwanted motions of the slave manipulators, thus autonomous risk detection methods robust to communication problems are desired. Hence, we propose to estimate the mechanical properties of the organs being manipulated by the slave manipulators using force sensors in order to detect the risk of organ damage. To validate this concept, we developed a system to autonomously detect the perforation risk of a membrane-like object by demonstration-based learning. We performed experiments to evaluate the developed system, and the results showed that the developed system successfully detected the risk of perforation.
