A flight computer that understands the flight
A rocket does not experience one steady environment. It sits motionless on the pad, accelerates violently through boost, crosses into coast, reaches apogee, and then swings beneath its parachutes. Vega does more than record those measurements: it continuously determines which part of the flight is happening and adapts its decisions accordingly.
Its state estimator runs 100 times per second. A Kalman filter combines fast accelerometer measurements with barometric altitude to estimate height above the launch site, vertical velocity, and accelerometer bias. During boost it reduces its trust in pressure measurements disturbed by airflow and transonic effects; during coast it gradually brings the barometer back into the calculation; under parachute it stops relying on misleading vertical acceleration.
Critical events are based on stable trends preventing anomalous sample from interfering with expected behavior. Sensor spikes are filtered before estimation, and conditions such as apogee or main deployment must remain valid before Vega acts.
