NEUROFEEL: SOFTWARE AND HARDWARE METHODS FOR PHYSICAL INTERACTION VIA EVENT-BASED TACTILE SENSOR

Abstract

Achieving the same or even superior accuracy in tasks performed by human hands is a paramount goal for machines [39]. To meet this demand, the development of tactile sensors that seamlessly adapt to diverse environments and conditions is essential. Drawing inspiration from the intricacies of the human tactile perception system, particularly the bio-inspired mechanoreceptors (both SA and FA), this paper introduces a novel approach by incorporating soft silicon fingertip with an event-based camera. In a pioneering endeavor, this optical sensor seamlessly integrates soft robotics, optical sensing, leveraging event-based vision, and the incorporation of whisker sens- ing. The optical sensing component offers heightened spatial resolution compared to traditional sensing systems, while the event-based nature of the camera employed in this sensor delivers an unprecedented temporal resolution—1,000 to 1,000,000 events per second [46]. This exceptional temporal resolution is particularly crucial for tasks such as texture detection. At the forefront of the sensor lies a silicon-made tip, mirroring the flexibility of human skin. This bio-inspired design ensures a tactile sensor with not only advanced optical capabilities but also a structure that emulates the adaptability and responsive- ness of the human touch. The sensor’s potential is further accentuated as it undergoes testing through a texture discrimination task. To facilitate this, a specialized algo- rithm, tailored to handle the event data output of the camera, will be implemented. This algorithm sets the groundwork for future advancements, with the intention of evolving into a sophisticated ML/DL algorithm. By integrating these bio-inspired principles, the proposed sensor endeavors to push the boundaries of tactile sensing, contributing to the development of machines capable of various types of interactions like texture recognition, slip detection, static and dynamic pressure sensing.