Force decoupling is essential for high-fidelity proprioception sensing in robotic systems. However, existing approaches rely on bulky multi-axis transducers or complex signal post-processing, hindering response speed and scalability. This work presents a force sensor employing a compliant mechanism-inspired meta-structure to mechanically decouple multidirectional forces. The metamaterial structure redistributes internal stiffness to channel force components into desired sensing elements while attenuating off-axis loads. This geometry-driven approach achieves high directional sensitivity (decoupling ratio >200) without complex signal processing. The design is sensing material-agnostic and integrates with capacitive, piezoresistive, and piezoelectric materials, enabling both static force measurement and high-bandwidth dynamic sensing (up to 1 kHz) within a unified architecture. Experimental results demonstrate accurate multidirectional force sensing enabling proprioception capabilities including texture recognition (95%+ accuracy), slip detection, and precise manipulation control in dexterous robotic hands. The sensor's compact, 3D-printable design supports scalable array integration, making it well-suited for next-generation robotic systems requiring rich tactile feedback.
Project currently funded by: Industry Sponsored Research