I. Introduction
Multicopters rely on robust and efficient trajectory planning for safe yet agile autonomous navigation in complex environments [1]–[6]. For robotics, precisely incorporating dynamics, smoothness, and safety are essential to generate high-quality motions. Moreover, lightweight robots, such as multicopters under size, weight, and power constraints, put further hard requirements on the real-time computing using limited onboard resources. Despite that various successful tools in general-purpose kinodynamic planning or optimal control have been presented, few of them guarantee efficient online planning while also considering general constraints on dynamics for multicopters. Consequently, existing applications often use oversimplified requirements on trajectories for better computation efficiency, thus limiting the full exploitation of vehicle’s capability.