Real-Time Nonlinear Characterization of Soft Tissue Mechanical Properties

Abstract

Online soft tissue characterization is important for robotic-assisted minimally invasive surgery to achieve precise and stable robotic control with haptic feedback. This paper presents a new nonlinear recursive adaptive filtering methodology for online nonlinear soft tissue characterization. An adaptive unscented Kalman filter is developed based on the Hunt-Crossley model by windowing approximation to online estimate system and measurement noise covariances. To improve the accuracy of noise covariance estimations, a recursive formulation is subsequently developed for estimation of system and measurement noise covariances by introducing a weighting factor. This weighting factor is further modified to accommodate noise statistics of large variation which could be caused by rupture events and geometric discontinuities in robotic-assisted surgery. Simulations, experiments, and comparison analyses demonstrate that the proposed nonlinear recursive adaptive filtering methodology can characterize soft tissue parameters in the presence of system or measurement noise statistics in both small and large variations for robotic-assisted surgery. The proposed methodology can effectively estimate soft tissue parameters under system and measurement noises in both small and large variations, leading to improved filtering accuracy and robustness in comparison with UKF.