The recent advancements in control systems for robotic platforms have shown a significant shift towards integrating learning-based approaches with traditional model predictive control (MPC) frameworks. This integration aims to enhance the adaptability and efficiency of control algorithms, particularly in dynamic and uncertain environments. The focus has been on developing lightweight, solver-aware learning models that can operate within the computational constraints of tiny robots, enabling high-rate control and improved tracking performance. Additionally, the use of transformer-based neural networks within the MPC optimization process has demonstrated substantial improvements in convergence rates and runtime, making it feasible for real-world applications. Notably, the field is also exploring distributed MPC strategies for multirotor systems, leveraging online learning to handle complex scenarios such as autonomous landings on moving platforms. These developments collectively push the boundaries of what is achievable with current robotic control technologies, paving the way for more autonomous and versatile robotic systems in various applications.