Adaptive Kinematic Model Learning for Macro-Micro Surgical Manipulator Control

Robot assisted surgery may have a huge role in providing treatments and guaranteeing safety of doctors and patients. However, surgical robots need accurate models and proper control strategies in order to perform the desired procedure safely and satisfy possible constraints, such as Remote Center of Motion. Generally, traditional rigid manipulators are used along with surgical robotic tools to augment the redundancies of the system and better solve constrained motion tasks. Here, we propose an adaptive modelling approach to build the forward kinematics of a surgical robot and a controller using the learned model in conjunction with a model of a serial-link manipulator, capable of ensuring accurate and correct task execution, while satisfying additional constraints. Simulated tests show the capabilities of our method of guaranteeing accurate execution both in case of full autonomy and in teleoperation, also with unpredicted changes in the robot model.