Autonomous robots and drones increasingly rely on intelligent sensor data processing to navigate and interact with complex environments. This chapter explores how machine learning (ML) empowers these systems to interpret multimodal sensor inputs including LiDAR, IMU, cameras, radar, and tactile sensors for tasks such as localization, obstacle avoidance, and decision-making. It covers the synergy between sensing modalities and ML algorithms, highlighting techniques like supervised learning, deep learning, and reinforcement learning for robust real-time performance. Challenges such as sensor noise, data heterogeneity, and synchronization are discussed, along with solutions like data preprocessing and sensor fusion. The chapter also examines advanced topics including representation learning, embodied intelligence with soft sensing, and edge-AI through TinyML. Case studies in UAV object detection, predictive maintenance in robots, and human-robot interaction illustrate practical implementations. Finally, it addresses emerging trends such as self-supervised and federated learning, quantum ML, and bio-inspired sensing systems. This comprehensive overview underscores the transformative role of ML in enhancing the autonomy, adaptability, and reliability of robotic and drone systems across domains like healthcare, manufacturing, logistics, and smart cities.