Dual-Modal Flexible Tactile Sensor for Complex Scenario Material Recognition

Abstract

Single-modal tactile sensors are well-suited for executing straightforward and efficient tasks in fixed scenarios. Nevertheless, the sensor's recognition accuracy degrades markedly upon exposure to complex environments. Accordingly, this paper proposes a dual-modal flexible tactile sensor (DFTS) tailored for material recognition in complex scenarios, delivering outstanding simultaneous dynamic and static perception and material cognition. In the triboelectric part, innovatively combining MXene with three-dimensional conductive sponges. Their synergistic operation amplifies the electrostatic-induction effect, markedly enhancing the triboelectric unit's output performance. In the piezoresistive part, the abrupt changes in resistance generated by the pyramidal microstructure achieve high sensitivity. Optimizing the MXene and PEDOT:PSS ratio delivers peak performance for the piezoresistive layer. Furthermore, vertically stacked yet electrically independent triboelectric and piezoresistive modules enable simultaneous detection of dynamic and static pressures, allowing the robot to perceive the entire grasp-and-release sequence in real time. Most importantly, an intelligent tactile perception system was constructed by integrating the VGG network model. Deployed in complex environments for material recognition, the tactile system achieved an accuracy of 98.77 %. The accuracy improved by approximately 11.55% comparing to single-modal tactile sensors. For the future, this system will demonstrate tremendous advantages in assistive devices for the visually impaired and bionic prosthetics.