Humanoid robots operating in unstructured environments require distributed tactile perception systems capable of maintaining sensing performance under complex mechanical deformation. Conventional electronic skin (e-skin) technologies have demonstrated promising capabilities in pressure sensing and human–robot interaction; however, their performance is often degraded by stretching, bending, and twisting deformation encountered in highly articulated robotic structures. Furthermore, most existing e-skin systems focus on low-level sensing and lack the intelligence required to transform tactile information into meaningful robotic actions.

This paper presents a conceptual framework for a multi-sensory auxetic electronic skin (e-skin) integrating stretchable printed circuits, multimodal tactile sensing, and Physical AI-based tactile intelligence for humanoid robots. The proposed architecture combines auxetic conductive interconnects, neutral-plane structural engineering, and a high-density sensor matrix capable of simultaneously detecting pressure, shear force, vibration, temperature, and proprioceptive deformation. The auxetic routing network enables enhanced mechanical compliance under stretching, bending, and twisting conditions, while maintaining electrical continuity and signal stability.

A hierarchical tactile intelligence framework is further proposed, integrating tactile signal acquisition, deep learning-based perception, semantic interpretation, and action generation. Through convolutional neural networks (CNNs) and transformer-based learning architectures, tactile patterns are transformed into high-level physical interaction representations that support object manipulation, slip detection, touch-guided learning, and human–robot collaboration.

The proposed framework establishes a scalable foundation for next-generation humanoid robotic skin systems and contributes toward the realization of embodied Physical AI, where robots learn and interact through distributed tactile perception rather than relying solely on visual information.