Three-Dimensional Ag2Se Thermoelectric Network Advances Multimodal Intelligent Perception

Embodied artificial intelligence (Embodied AI) depends on the integration of multimodal sensing and physical interaction to achieve autonomous perception and adaptive control in complex environments. However, conventional multimodal sensing systems often rely on separate sensors for each physical stimulus, leading to integration challenges and signal interference. Here, we report a monolithic bimodal sensor based on a 3D Ag 2 Se thermoelectric network that enables simultaneous and decoupled detection of temperature and pressure within a single device. The sensor demonstrates a high temperature sensitivity of −122.7 µV K −1 with a fast response time of ∼0.14 s, and a high pressure sensitivity of −2.94% kPa −1 with a fast response time of about 0.08 s, indicating excellent signal isolation between the two sensing channels. When combined with deep learning, the sensor enables a real-time intelligent sorting system that achieves 96% recognition accuracy across nine materials, outperforming single-mode sensing approaches. This work provides an integrated multimodal sensing solution that advances the deployment of embodied AI in complex environments.