Two Layer Sliding Mode-Model Predictive Control Scheme For Image-guided Robotic Flexible Needle Steering

This paper presents a two-layer sliding mode model predictive control (SMPC) for steering needle in 3D environment. These two layers are applied with two different sampling timescales and two different models. In this scheme high-level controller uses model predictive control to calculate feasible optimum needle base angle to reach the target with minimum rotation and needle length while avoiding obstacles using artificial potential field (APF). The MPC approach proposed on a cost function that satisfies non-holonomic needle constraints. Simplified discrete model is developed to reduce high-level controller computational difficulties. Unicycle 3D needle steering model is used to provide stability of low-level sliding mode controller. Finally, regulation and trajectory tracking problems are examined to illustrate performance of the proposed method.