Rapidly deployable surgical robots pose minimal interruption to surgical workflow and require minimal setup time and equipment to support deployment. This paper explores the concept of rapid deployment through the use of in-vivo sensory information to adapt a pre-operative surgical plan and to increase robustness against registration and misalignment errors during robot deployment. Robotic insertion of cochlear implant electrode arrays is presented as a benchmark application demonstrating this concept. Two key ideas are presented within the context of this application: First, a hybrid position and admittance controller is used to define an insertion path plan that is modified based on in-vivo force measurements in order to reduce sensitivity to misalignment errors. Secondly, a new concept allowing the use of force cues to determine the onset of advance-off stylet electrode array insertion is presented. The new controller is tested with electrode insertions in both plastic models and human cadaveric specimens. The experiments show that insertion forces may be maintained or reduced compared to preplanned trajectories relying solely on the initial registration.

Perimodiolar electrode arrays for cochlear implant surgery present surgeons with challenges during their insertion into the cochlea. These challenges stem from the inherent flexibility of these arrays and the fragile intracochlear anatomy and has opened opportunities for high precision robotic systems. This paper investigates effects of insertion speed on insertion forces using perimodiolar electrode arrays in human temporal bones and complements previously published results on insertion force and speed relationships when using straight outer-wall electrode arrays. The paper presents a custom-made parallel robot equipped with a high resolution force/moment sensor as an insertion platform. Experimental results using 42 insertions in 8 temporal bones are analyzed and grouped by insertion speed. Our results show a reduction in insertion forces up to insertion speeds of 1 mm/s. The second aspect of this paper is the investigation of the feasibility of using insertion force and speed data to sense conditions predicting the onset of electrode tip fold over, a complication resulting in poor placement of electrode contacts. It is shown that tip folding can be detected by features taken from insertion force data when employing a support vector machine classifier. The capability to detect such errors in a procedure is a critical component to the development of a practical surgical tool.

This paper evaluates the potential for using robotic assistance for the insertion of commercial Perimodiolar Electrode Arrays (PEA's) that use a stylet and a pre-curved electrode body.We propose an algorithm for coordinated robotic insertion of the pre-curved electrode body and pulling of the stylet in order to provide partial control over the PEA shape. However, for robotic insertion to be a viable option for PEA's, an understanding of the statistical variability of their characteristics is essential. This paper investigates the kinematics of PEA's, their shape variability, and their expected performance during robotic insertion. An investigation of the required number of Degrees of Freedom (DoF) and the workspace is presented. A study of shape variability is carried out for characterizing shape repeatability across different insertion trials using the same electrode array. Also a study of shape variability is carried out between different electrode arrays. We simulate and evaluate the expected performance of three and four DoF robotic instruments for PEA insertion. The results support our proposed algorithm for coordinated insertion of the electrode array body and pulling of the stylet. They show that existing commerical perimodiolar arrays provide good shape repeatability. These results also point to the need for designing robotic insertion tools with at least four DoF. We believe that these results provide guidelines for future robotic instrument design for cochlear implant PEA's.