Particle filtering based gyroscope fault and attitude estimation with uncertain dynamics fusing camera information

The problem of gyroscope faults and a rigid body attitude estimation with an uncertain dynamic model are presented in this paper. Gyroscope erroneous and faulty measurements are fused with Photogrammetric camera measurements. The attitude dynamics in both parameterized and non-parameterized forms is considered where the tensor of inertia is the unknown parameter in the parameterized form. Angular maneuvers are modeled by considering angular acceleration as an unknown input to angular velocity dynamics. It is modeled by Wiener process and also augmented to the state parameters to be estimated. Sensory biases and drifts are augmented to the attitude state parameters. The unknown tensor of inertia is estimated using particle filtering (PF) based method leading to an adaptive approach. Having imperfect attitude rate dynamics due to the lack of exact knowledge of inertia tensor, a modified particle filtering (MPF) approach is proposed for attitude estimation. The main idea behind the MPF is to engage both system and measurement models in particle generation. Experimental results based on data from a 3D micro electro mechanical system inertial measurement model (MEMS IMU) and a 3D camera system are used to demonstrate the efficiency of the method.