Efficient covariance intersection of attitude estimates using a local-error representation

An approach to overcome the difficulties in the covariance intersection (CI) fusion of attitude estimates by using a local MRP (modified Rodrigues parameters) approach that circumvents the quaternion norm constraint while maintaining a singular free attitude representation was studied. The increased numerical conditioning of the local-error MRP algorithm means that the results are less susceptible to round-off errors than existing algorithms. The mathematics of the problem were shown to be such that a minimal number of computationally expensive matrix inverses are required, meaning that little processing power is required for implementation of the algorithm.