Motion Compensation in Minimally Invasive Robotic Surgery

Minimally invasive beating heart surgery allows a very sparing operation, but increases the requirements for the surgeon: The remaining motion of the mechanically stabilized beating heart makes fast and safe surgery difficult. The goal of an advanced robotic surgery system is to compensate for this motion. This work presents control and vision algorithms necessary for such novel robotic surgery applications. Cartesian position and velocity control laws in a minimally invasive surgery environment allow correct hand eye coordination which the surgeon got used to in open surgery. In combination with appropriate filtering and scaling of input commands, high accuracy manipulation of fine structures is possible. A self-adapting Cartesian force control law reduces the risk of unintentional damage of delicate tissue structures. Motion of the mechanically stabilized beating heart is locally captured by tracking natural landmarks with an affine motion model. To circumvent disturbances of the tracking approach, specular reflections on the heart surface have to be handled appropriately. Automatic detection of landmarks allowing reliable determination of the affine model parameters can be achieved by special confidence measures. A new motion prediction framework is introduced to further increase robustness of the motion tracking scheme. This framework is able to compensate for short occlusions and small disturbances. Additional signals correlated with the heart motion (e.g. electrocardiogram) are included in this prediction scheme to reduce the dependency on visual information.