A control-physics informed machine learning (CPhy-ML) framework is introduced to infer drone intentions by integrating deep learning with aerospace physics. This approach improves the accuracy and stability of drone intention classification and prediction, outperforming traditional methods. The framework combines data-driven learning with physics-based models to reduce bias and enhance generalization. It uses a hybrid classifier with attention mechanisms and deep LSTM autoencoders for trajectory intention classification, achieving high accuracy and robustness. The framework also incorporates multiple experts for trajectory intention regression, improving prediction accuracy by capturing trajectory variability. Physics-informed models are used to stabilize predictions and enhance robustness, particularly for long-term forecasts. The framework also infers reward functions to understand drone control objectives, using model-based reward-shaping inverse reinforcement learning. The CPhy-ML framework demonstrates superior performance in trajectory prediction and intention inference, providing insights into drone behavior and enabling more effective countermeasures against malicious activities. The framework is evaluated using synthetic and real-world data, showing improved accuracy and robustness compared to conventional methods. The framework's performance is further enhanced by incorporating control inputs and physics-informed models, leading to more accurate predictions and better understanding of drone intentions. The framework is limited by data variability and richness, requiring more diverse data for accurate generalization. Overall, the CPhy-ML framework offers a robust and effective approach to drone intention inference, with potential applications in counter-drone technologies and autonomous systems.A control-physics informed machine learning (CPhy-ML) framework is introduced to infer drone intentions by integrating deep learning with aerospace physics. This approach improves the accuracy and stability of drone intention classification and prediction, outperforming traditional methods. The framework combines data-driven learning with physics-based models to reduce bias and enhance generalization. It uses a hybrid classifier with attention mechanisms and deep LSTM autoencoders for trajectory intention classification, achieving high accuracy and robustness. The framework also incorporates multiple experts for trajectory intention regression, improving prediction accuracy by capturing trajectory variability. Physics-informed models are used to stabilize predictions and enhance robustness, particularly for long-term forecasts. The framework also infers reward functions to understand drone control objectives, using model-based reward-shaping inverse reinforcement learning. The CPhy-ML framework demonstrates superior performance in trajectory prediction and intention inference, providing insights into drone behavior and enabling more effective countermeasures against malicious activities. The framework is evaluated using synthetic and real-world data, showing improved accuracy and robustness compared to conventional methods. The framework's performance is further enhanced by incorporating control inputs and physics-informed models, leading to more accurate predictions and better understanding of drone intentions. The framework is limited by data variability and richness, requiring more diverse data for accurate generalization. Overall, the CPhy-ML framework offers a robust and effective approach to drone intention inference, with potential applications in counter-drone technologies and autonomous systems.