Project Summary/Abstract In beating heart procedures such as tricuspid annulus reduction, only visual feedback is provided to the surgeon via imaging and the accuracy in operating surgical tools relies on the quality of the intraoperative imaging. Hence, there is a need for safe, advanced methods to monitor position and interaction forces of the robot with the heart tissues to ensure safe and accurate manipulation of delicate heart structures. This proposal aims at developing an exteroceptive probe that detects pressures applied to the heart structures by the soft robot developed in the parent grant and contact forces within the heart. Such sensing capabilities will complement conventional imaging in increasing accuracy in positioning the robot inside the beating heart. It will also allow to monitor interaction forces and contacts with the heart anatomy. Machine learning (ML) methods will be investigated to discriminate and decouple the various raw sensory inputs and use them to identify tissues the probe is in contact with for better localization, safety, and to guide the procedure. The proposed research has two aims: Aim 3 focuses on sensor design and fabrication of the probe; Aim 4 handles sensor integration with the soft deployable robot, sensory interpretation based on ML, and function validation through in-vitro and ex-vivo testing. Combined with external and internal imaging, we hypothesize that the addition of this sensor will increase the accuracy and safety during beating heart surgical procedures. To monitor pressures applied distally by the robot a soft pressure sensor will be developed. Miniaturized soft contact sensors will also be integrated on the robot. These sensors will be inspired by the hair-like cilia observed in biology. Acting similarly to cantilever beams, cilia-like extensions can deflect via fluidic drag forces producing piezoelectric charges and generating a direction-dependent potential to sense flow rate and direction. We will develop cilia-inspired sensors around the pressure sensor to monitor contacts experienced by the robot. Taking advantage of the multiple sensing modalities on the exteroceptive prove, we will investigate the use of ML application to enhance soft robot awareness and provide tissue recognition and haptic feedback to the surgeon. These additional components will improve the robot perception of the surroundings. This will lead to 1) increased accuracy in positioning by integrating the information from onboard sensing with external imaging, 2) increase safety via monitoring contact and exchanged forces with surrounding heart structures and potentially paving the way towards integration of haptic feedback, and 3) further increase the accuracy in the control of the robot by exploiting sensor data to correct model estimation errors. 1