速度提升447%！MIT成果登《Science》子刊，算法解锁微型机器人“昆虫级”敏捷飞行

Core Viewpoint - MIT scientists have successfully developed a lightweight flapping robot weighing only 750 milligrams that can replicate the agile flight patterns of fruit flies, demonstrating remarkable capabilities such as performing ten consecutive rolls in 11 seconds and maintaining stability in strong winds [1][2]. Group 1: Challenges in Micro-Robotics - Researchers face a "trilemma" in achieving a balance between agility, robustness, and computational efficiency in micro-robotics, which has historically limited performance [4][5]. - Mid-scale flying robots often sacrifice agility for better computational and load capabilities, while insect-sized robots have had to limit speed and movement to maintain stability, resulting in subpar performance [5][6]. Group 2: Breakthroughs in Control Algorithms - The MIT team innovated by overhauling the control architecture rather than adding hardware, creating a robust tube model predictive control (RTMPC) framework that allows the micro-robot to perform insect-like maneuvers [7][9]. - The control system mimics a "master-apprentice" training process, where a high-performance RTMPC plans optimal flight paths, while a lightweight neural network learns from the expert's decisions for real-time control [10]. Group 3: Performance Achievements - The micro-robot achieved significant breakthroughs in speed and agility, reaching a maximum speed of 124 cm/s and lateral acceleration of 11.4 m/s², with improvements of 245%, 182%, and 243% in acceleration, deflection angle, and speed compared to previous studies [12][13]. - It demonstrated robust performance under disturbances, maintaining a position error of only 4.72 cm despite a 33% mapping error and a 4.58 cm error in strong winds, showcasing its stability [13][15]. - The robot set new records for tracking complex trajectories, achieving speeds of 197 cm/s on an "8" shaped path, a 446% increase from prior records [15][17]. Group 4: Future Pathways - The research team acknowledges challenges in transitioning from laboratory demonstrations to real-world applications, particularly in developing high-density micro-batteries and efficient wireless communication modules for autonomous flight [19]. - Future iterations of the micro-robot will require integrated sensors and embedded state estimation algorithms to operate independently, akin to functioning without external guidance [19][21]. - The study highlights the potential for scalable computational efficiency, allowing for significant reductions in neural network size while maintaining stable flight, paving the way for advanced micro-robotics [19][22].