Distributed Force Sensing on Large-Area E-Skin Inspired by Human’s Visual and Tactile Synergy

Robots always perceive and produce multiple forces simultaneously at distributed points on their skin during their interaction with the environment. Traditional fiber Bragg grating (FBG) distributed force sensors assembled on robots rely on the coupling performance of multisegments of the FBG, which makes them quite fragile and can acquire force information at just one single point. Therefore, in order to achieve an effective distributed force sensing, inspired by the synergistic effect of the human visual and tactile senses, an FBG-VISION sensor is specially designed, and a YOLOv8-RFR-SVM (Y-R-S) algorithm is proposed in this article. The FBG-VISION sensor combines a visual sensor with an FBG force sensor. The Y-R-S is composed of YOLOV8, random forest regression (RFR), and a support vector machine in an orderly manner. This comprehensive combination is able to overcome the multiforce detecting problem in the distributed force sensing and effectively recognizes the position and magnitude of multipoint forces. In order to verify our proposed sensor and the corresponding Y-R-S, two experiments, namely a single-point experiment and a multipoint experiment, are conducted. In the single-point experiment, we made a comparison with other sensing approaches we could find, indicating that the prediction accuracy of both position and force with our proposed method is superior to those methods. The multipoint experiment proved that our proposed method can effectively measure the force information at multiple points.