Intelligent Range Prediction and Trip Planning System for Amazon Electric Boats

Abstract

The transition to sustainable transportation is critical in the Amazon, where riverine routes are vital for mobility and commerce. The adoption of electric boats is hindered by limited charging infrastructure and unpredictable environmental conditions. To address this challenge, we present an integrated system that combines real-time sensor monitoring with machine learning-based range prediction to reduce operator range anxiety and enable intelligent trip planning. The system incorporates battery discharge modeling based on Peukert's Law and employs Gaussian Process Regression, which achieved superior performance (R² = 0.957, RMSE = 2.545) among evaluated methods through its balance of accuracy, computational efficiency, and uncertainty quantification. The system was validated on an electric boat operating on the Xingu River, Brazil, equipped with LiFePO4 batteries. The system provides boat operators with real-time dashboards displaying live navigation data, battery status, and reliable range estimates, significantly reducing operational uncertainty. This work demonstrates a practical solution for accurate range prediction in remote environments, advancing sustainable electric maritime mobility (green computing) in the Amazon, which can be applied to similar waterways worldwide.