Stanford autonomous helicopter dataset and the HFDS-100 real-flight dataset

The Stanford Autonomous Helicopter Dataset captures a series of high-agility aerobatic flight demonstrations performed by an instrumented radio-controlled helicopter in 2008. The platform executes a wide range of aggressive maneuvers\u2014including flips, rolls, stall-turns, pirouettes, chaos maneuvers, and full aerobatic routines\u2014under high-bandwidth state observation. The dataset records 9-dimensional flight states (3-axis velocity, 3-axis angular velocity, and Euler angles) and 4-dimensional control inputs (aileron, elevator, rudder, and collective), sampled at high frequency with non-uniform timing characteristics typical of real flight experiments. These trajectories span both nominal and highly nonlinear dynamical regimes, exhibiting substantial coupling, rapid transients, and multi-regime behaviors.