A Tapered Whisker-Based Physical Reservoir Computing System for Mobile Robot Terrain Identification in Unstructured Environments

Abstract

This paper presents a physical reservoir computing system based on a tapered robotic whisker for terrain classification and roughness estimation. The method uses vibration signals generated by ground interaction and maps them into useful features through the whisker’s nonlinear dynamics. The results show that morphology-aware sensing can support lightweight, real-time terrain perception for mobile robots.

Citation

Yu, Z., Perera, S., Hauser, H., Childs, P., & Nanayakkara, T. (2022). A Tapered Whisker-Based Physical Reservoir Computing System for Mobile Robot Terrain Identification in Unstructured Environments. IEEE Robotics and Automation Letters, 7(2), 3608-3615. https://doi.org/10.1109/LRA.2022.3146602