Step-Ahead Forecasting of Wireless Signal Strength for Industrial AMRs: A Deep Learning Approach with LSTM and GRU Models

Accurate Wi-Fi signal strength prediction is fundamental for Autonomous Mobile Robots (AMRs) that require uninterrupted connectivity in industrial environments. This study presents a deep learning framework for forecasting the Received Signal Strength Indicator (RSSI) using recurrent neural networks (RNN). To this end, a novel dataset comprising 85,000 samples was collected by recording telemetry and RSSI data from an AMR operating in a real-world metal-mechanics industrial plant. Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models were systematically optimized and evaluated. The results demonstrate that the proposed approach successfully learns the complex spatio-temporal signal patterns, achieving high accuracy and reliable forecasts for multiple steps ahead within a practical time horizon for real-time decision-making. This forecasting capability enables AMRs to proactively manage connectivity by anticipating communication failures to trigger handovers or recalculate routes. The framework, therefore, represents a step toward enhancing the operational robustness and safety of mobile robots in dynamic industrial environments. The dataset and source code are made publicly available to foster reproducibility and future research.