The decline in battery voltage during unmanned aerial vehicle (UAV) flights poses a critical challenge that can compromise operational safety, reduce flight time, and limit mission efficiency. Battery degradation over time leads to poor motor performance, increases the risk of mid-air failure, and elevates maintenance costs. This study proposes the development of a multivariate time series forecasting model to estimate the Remaining Useful Life (RUL) of drone batteries, integrating various environmental and operational variables such as voltage, temperature, wind speed, and flight duration. To enhance predictive accuracy, machine learning techniques such as Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Extreme Gradient Boosting (XGBoost) are explored. The methodology follows a structured process including data collection, Exploratory Data Analysis (EDA), correlation analysis, and model training. Data preprocessing techniques like feature engineering and imputation are used to manage noise and missing values. The final model is deployed in an interactive software tool that allows for real-time predictions, proactive maintenance, and enhanced mission planning. This approach contributes to the field by improving safety, reducing operational costs, and promoting sustainability in drone applications across delivery, surveillance, and agriculture. The findings are expected to advance battery health monitoring practices and support broader use of predictive analytics in drone operations.