Surgical tools pose estimation for a multimodal HMI of a surgical robotic assistant

The main objective of this paper is to minimize the occluded areas in order to recognize the navigation of the surgeon's tools for a two-arm autonomous robotic system for laparoscopic procedures. This robotic assistant needs the tracking of the surgeon's surgical gestures in order to recognize the current maneuver and to execute the automated tasks of the robot. The surgical tools pose estimation is carried out by a Multiple Extended Kalman Filter (MEKF), where the movement models of the surgical tools depend on the maneuver which is being developed. This information is obtained by a maneuvers recognition system which is a part of the multimodal human machine interface (HMI) of the robot. The method proposed for reducing shadows has been applied to three in-vitro maneuvers which appear in the majority of the surgical protocols. The experiments show the behavior of this method for different time intervals of the occlusions.