Multimodal Perception-Driven Optimization of Mobile Interaction Interfaces and Supply Chain Efficiency

Abstract

In mobile supply chain operations, constraints imposed by mobile terminal resources, fluctuations in operator cognitive load, and inadequate interface adaptability severely limit operational efficiency and increase the risk of operational errors. To address these challenges, a mobile interaction interface optimization and efficiency enhancement model was proposed, grounded in cloud-edge collaborative multimodal perception. A lightweight temporal fusion network, integrating time-division cross-modal attention and knowledge distillation, was developed to enable efficient processing of multimodal data at the edge and accurate prediction of operator intent, thereby accommodating the resource constraints of mobile devices. A mechanism for estimating instantaneous cognitive load, driven by multimodal proxy indicators, was constructed. Combined with meta-reinforcement learning, a load-aware dynamic interface reconfiguration strategy was formulated, enabling real-time adaptation between the interface and operator state. A multimodal intent-driven Markov decision process-based workflow preloading model was established, and an online Bayesian optimization framework was incorporated to form a closed-loop optimization system, effectively improving supply chain efficiency. This study provides an innovative solution for the application of mobile interaction technologies in supply chain scenarios and promotes the deep integration of mobile multimodal interaction with industrial environments.