Estimation of accelerometers' bias in decentralized relative localization via the rotating UWB tag

This paper proposes a novel algorithm for cooperative relative localization of multiple aerial robots. The algorithm estimates acceleration biases and velocities alongside relative positions, leveraging relative range and inertial measurements. A group of quadrotors, consisting of one master and two regular units, collaboratively estimates the relative positions. The master quadrotor estimates the acceleration biases of each unit and shares these values to compensate for their effects on position estimation. The proposed method incorporates geometric constraints as additional measurements, enabling accurate estimation. When applied individually, the law of cosines and the velocity constraint reduce the average estimation error of acceleration biases by 71 % and 81 %, respectively. An observability analysis is conducted, providing necessary conditions for system observability. The performance of the proposed method is evaluated through simulations under various conditions-including noise-free and noisy measurements, as well as uncertainties in initial estimation guesses-and through experimental validation. Comparative analysis with existing approaches demonstrates that the proposed method effectively estimates velocities by compensating for acceleration biases, with the potential to enhance absolute navigation accuracy.