Frame‐Channel Engineered Multimodal Transistor Integrating Neuromorphic Computing with Cross‐Modal Edge Intelligence
Bo He, Bin Bao, Rodrigo Martins, Shouguo Wang, Gang He
Abstract
Abstract The sustained advancement of neuromorphic electronics aims to design electronic systems that emulate biological neural networks, addressing the hardware computational demands of supercomputers and the development of multimodal edge devices in the context of artificial intelligence. However, a critical challenge in current neuromorphic devices replicating synaptic functions lies in their overreliance on non‐volatile conductance updates for simulating neuromorphic behaviors, which restricts higher‐order biomimetic capabilities and impedes modality expansion. Here, a transistor active‐layer architecture is proposed featuring a frame‐channel (FC) design, enabling full‐dimensional spontaneous interactive coupling between core materials and physical spatial signals. This innovation facilitates the first bio‐inspired inhibition‐integrated regulatory mechanism in optoelectronic synapses, demonstrating dual‐mode neuromorphic functionalities: tripartite synaptic modulation and lateral inhibition‐based visual processing. Importantly, the device can cross‐modally switch to an intelligent respiratory signal sensor, integrated with a lightweight algorithm to construct an efficient interactive edge system. This synergistic strategy, merging material architecture, neuromorphic dynamics, and algorithmic design, establishes a paradigm for developing smart functional electronics.