GPS Denied Localization and Magnetometer-Free Yaw Estimation for Multi-rotor UAVs

This paper presents a range-based localization scheme for multi-rotor systems in GPS denied environments and proposes a novel methodology to estimate the yaw attitude of the UAV during flight. The attitude and position are estimated by fusing the accelerometer, gyroscope, and range information using Extended Information Filter (EIF). The heading is estimated without using the magnetic sensors. All family of Gaussian filters requires the correct noise parameters for convergence and accurate estimation. An optimization technique is used for tuning the estimator's parameters. Particle-Swarm Optimization (PSO) method is used for tuning the noise (covariance matrices) in the filter with the aid of ground truth in the initial flight. The effectiveness of tuned EIF is validated on the quadcopter platform with different environments, which shows superior performance compared to the manually tuned noise parameters. The magnetometer free yaw attitude estimation is of great significance in environment having high magnetic deviation.