Model-Based Haptic Guidance in Surgical Skill Improvement

This paper defines a model-based control hap tic guidance (MPC-HG) approach to improve the performance of a user in a minimally invasive surgery (MIS) training while performing a surgical related task. In this approach, a robot applies controlled forces on the hand of the user to guide him/her through an MIS training task according to an MIS reference model and desired set of motions. The main challenges for such physical human-robot interactions (pHRI) involve precise modeling of human motion to develop an expert model and invoking adaptive control approaches for force rendering. Thus, a hidden Markov model (HMM) and two control strategies are presented and evaluated. The model automatically creates a set of states and is trained accordingly based on a reference set of motion in the form of robot Learning from Demonstration (LfD). The control strategies include a novel model predictive control (MPC) framework and classical spring impedance controllers. Furthermore, the effects of hap tic guidance (HG) on the performance of users have experimentally been investigated. The results confirm that HG is effective in improving user performance for a sample surgical related task in terms of time related metrics. The proposed MPC framework demonstrates promising results compared to the other control modes considered in the study.