Journal

Flew the DRAGON transition today. The gain-scheduled LQR held through the corridor — first clean fixed-wing to multirotor without any manual intervention. The altitude loss was under 4 meters which is well within what we planned for. Hard to describe what it feels like to watch something you spent months on just work.

Been thinking about the gap between simulation and hardware lately. The 6-DOF model matches flight data well on attitude but there is still something off in the translational response at high roll angles. My instinct is the propeller inflow model — BEM assumes uniform inflow and that breaks down when the disk is tilted. Worth revisiting the skewed wake correction.