Model predictive control architectures with force feedback for robotic-assisted beating heart surgery

Abstract

Minimally invasive surgery (MIS) offers considerable advantages for patients, lowering infection risks, and reducing trauma and convalescence times. Dedicated surgical robots significantly improve surgeon's skills especially for tasks requiring high precision such as lead placement and suturing. However, these robotic setups do not allow yet beating heart surgery with motion compensation functionalities. This paper tackles autonomous heart motion compensation with force feedback. We propose a cascade model predictive control (MPC) architecture with a Kalman Active Observer (AOB) in the loop, and compare it with the classical MPC approach. The cascade MPC-AOB control architecture has two loops. The inner one performs model-reference adaptive control, guaranteeing a desired force tracking dynamics. The outer one generates control actions to compensate physiological motions. Both MPC-based architectures are analyzed and experimentally evaluated. Two robots are used. A lightweight 4-DoF surgical robot generates desired surgical forces and a 3-DoF robot equipped with an ex vivo heart at the end-effector reproduces realistic heart motions.