Abstract
The goal of this thesis is the study and development of a tool for robot-driven adaptive needle insertion planning in order to assist percutaneous insertions of beveled semi-rigid needles guided by CT or MR images. The approach is based on a novel needle deflection prediction model that provides offline predictions and improve online the needle insertion performance. This improvement allows to compensate for environment uncertainties and approximations of needle-tissue interactions modeling. Results obtained from in vitro tests, using different robot platforms confirmed the viability of our method.