Constrained Optimal Control for Semi-Autonomous Robotic Systems

An autonomous robotic system can make decisions and perform actions without direct control, handling real-world conditions that are unpredictable and dynamic. In recent years, the research interest in robotic-assisted minimally invasive surgery is shifting from teleoperated devices to autonomous support systems for the execution of repetitive surgical steps (i.e., suturing, ablation and microscopic image scanning). This thesis aims to investigate the design of a supervisory controller for a semi-autonomous robotic system and the application of optimal control techniques to manage unexpected events and constraints during the execution of surgery. The management of such events will be demanded by the surgeon through a novel formulation of constrained bilateral teleoperation. For this reason, particular focus will be given to addressing MPC problems under input-output and state constraints for linear, non-linear and hybrid (i.e., with both continuous and discrete dynamics) systems. These constraints will be used to guarantee stability and enforce safety conditions during the teleoperated intervention. The thesis will also cover the research activities necessary to develop the EU funded H2020 SARAS semi-autonomous system. Such activities include the definition of a shared reference system for the multi-arm robotic setup and the registration of 3D preoperative medical data to the patient's anatomy.