PSO Aided Adaptive Complementary Filter for Attitude Estimation

Attitude estimation is one of the core frame- works used for navigating an unmanned aerial vehicle from one place to the other. This paper presents an Euler-based non-linear complementary filter (CF) whose gain parameters are obtained using particle swarm optimization (PSO) technique. It relieves the user from feeding the KP and KI parameters manually and adjust these parameters automatically when the error between the attitude measured from accelerometer and the CF increases above a particular threshold. The measurement unit for this research consists of micro-electro-mechanical-systems (MEMS) based low cost tri-axial rate gyros, accelerometers and magnetometers, without resorting to global positioning system (GPS) data. The efficiency of the CF is experimentally investigated with the help of reference attitude and the raw sensor data obtained from commercial inertial measurement unit (IMU). Simulation results based on the test data show that the proposed PSO aided non-linear complementary filter (PNCF) can automatically obtain the required gain parameters and exhibits promising performance for attitude estimation.