Abstract
Soft robotic modules have potential use for therapeutic and educational purposes. To do so, they need to be safe, soft, smart, and customizable to serve individuals' different preferences and personalities. A safe modular robotic product made of soft materials, particularly silicon, programmed by artificial intelligence algorithms and developed via additive manufacturing would be promising. This study focuses on the safe tactile interaction between humans and robots by means of soft material characteristics for translating physical communication to auditory. The embedded vibratory sensors used to stimulate touch senses transmitted through soft materials are presented. The soft module was developed and verified successfully to react to three different patterns of human-robot contact, particularly users' touches, and then communicate the type of contact with sound. The study develops and verifies a model that can classify different tactile gestures via machine learning algorithms for safe human-robot physical interaction. The system accurately recognizes the gestures and shapes of three-dimensional (3D) printed soft modules. The gestures used for the experiment are the three most common, including slapping, squeezing, and tickling. The model builds on the concept of how safe human-robot physical interactions could help with cognitive and behavioral communication. In this context, the ability to measure, classify, and reflect the behavior of soft materials in robotic modules represents a prerequisite for endowing robotic materials in additive manufacturing for safe interaction with humans.