Cascade robot force control architecture for autonomous beating heart motion compensation with model predictive control and active observer

Nowadays, robotic-assisted surgery does not allow beating heart surgery with autonomous motion compensation functionalities. This paper tackles this problem, based on a robotic control architecture that relies on force feedback. The algorithm merges two cascade loops. The inner one is based on the Kalman active observer (AOB), performing model-reference adaptive control to impose a well-defined stable plant. The outer one, based on a model predictive control (MPC) approach, generates control references for beating heart motion compensation. Two robots are used in the experiments. 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 motion. Additionally, robustness to cardiac stiffness mismatches is analyzed.