Drones require effective battery management, especially when using lithium-polymer (Li-Po) batteries, which are favored due to their high energy density and portability. Using NASA’s lithium-ion (Li-ion) battery datasets, this research develops a hybrid machine learning framework that can be used to Li-Po batteries to solve the problem of forecasting battery performance indicators like State of Charge (SoC) and State of Health (SoH). The approach combines an interactive Dash-based dashboard for battery performance visualization with ARIMA for time-series forecasting, LSTM for sequence modeling, and reinforcement learning for dynamic optimization. The study demonstrates how well the ARIMA and LSTM models can forecast battery behavior, with ARIMA demonstrating short-term accuracy and LSTM demonstrating long-term trend identification. Additionally, it shows how effective reinforcement learning is at maximizing energy use, particularly in dynamic drone environments. In order to improve practical application and refine forecasts, future research will concentrate on integrating real-world drone operating conditions and comprehensive Li-Po battery properties. This project demonstrates the framework’s scalability and adaptability, providing a solid means of enhancing real-time performance and precisely forecasting battery health, which makes it extremely useful for drone applications.