Sensor-based complementary globally asymptotically stable filters for attitude estimation

This paper presents the design, analysis, and performance evaluation of a new class of globally asymptotically stable filters for attitude estimation. The design is based directly on the sensor measurements as opposed to traditional solutions that resort to parameterizations of the attitude, e.g., Euler angles, quaternions, or rotation matrices, and therefore it does not have problems such as singularities, unwinding phenomena, or topological limitations for achieving global asymptotic stability. The proposed solution includes the estimation of gyros biases, incomplete sensor measurements, systematic tuning procedures, and also allows for the inclusion of frequency weights to model colored noise on the different sensing devices. Finally, and due to the inherent structure, the filters are complementary. Simulation results are included that illustrate the achievable performance in the presence of realistic measurements provided by a low-cost, low-power sensor suite.