Model Averaging and Input Transformation for 3D Needle Steering

Abstract

In this paper, a duty-cycle-based method is proposed for needle steering in 3D. The paper models the continuous 3D needle steering problem as a four-mode switching system and provides a new average-based formulation to transform the continuous input into a switching sequence. In this structure, the needle tip deflection control system is decomposed to two different 2D subsystems. Each subsystem has its own input, for which a controller can be designed to adjust a switching duty cycle. The duty cycles from the two subsystems are then combined to provide an axial needle rotation command to control the needle deflection in 3D. In order to show the application of the proposed formulation, robust sliding mode technique is employed to design controllers for each subsystem and, thus for the total system in 3D. The controllers are designed to be robust with respect to uncertainties in the value of the needle path curvature and to deal with measurement limitations. The performance of the proposed framework is shown by performing experiments in different scenarios.