Fast Adaptive Complementary Filter for Quadrotor Attitude Estimation During Aggressive Maneuvers

It is challenging to provide accurate attitude estimation for a quadrotor during aggressive maneuvers, which involve violent and fast motion with uncertainty on sensors. In this unique case, we propose a fast adaptive Complementary Filter, which fuses raw measurements from a MARG sensor system and could cope with the large range of quadrotor manuvers. The performance of the filter is validated on an actual quadrotor platform, by comparing with the Extended Kalman Filter (EKF) and the Error State Kalman Filter (ESKF) simultaneously. The experimental results show that the proposed CF has better estimate accuracy, low time-latency and stable orientation outputs in a quadrotor's attitude estimate.