Inexpensive Cubesat Attitude Estimation Using Quaternions and Unscented Kalman Filtering

This paper describes a quaternion implementation of an Unscented Kalman Filter for attitude estimation on CubeSats using measurements of a sun vector, a magnetic field vector and angular velocity. For faster convergence of the attitude estimate, a SVD-method solving Wahba’s problem has been proposed, which provides an initial attitude estimate. Using unit quaternions provides a singularity free attitude parameterization. However, the unity constraint requires a redesign of the Unscented Kalman Filter. Therefore, a quaternion error state is introduced. Emphasis has been put in making the implementation accessible to other CubeSat developers via pseudo code and simulations have shown that the extra computational cost of estimating bias in measurements is worthwhile. The simulations where performed in a simulation environment for the CubeSat AAUSAT3, where robustness has been an important factor during tuning of the attitude estimators. The results indicate that it is possible to achieve acceptable CubeSat attitude estimation, even during eclipse, on a limited budget without expensive high precision sensor setups.