Rotation vector-based attitude updating algorithms have been used as the mainstream attitude computation algorithms for many years. The most popular methodology for designing the rotation vector algorithm is by leveraging multiple samples of gyro integrated angular rate measurements. However, it has been pointed out by many researchers that the attitude updating accuracy is limited when using the multiple samples rotation vector algorithms, especially when the platforms work under high rate manoeuvres. The third-, fourth-, fifth- and sixth-order Picard component solutions of the rotation vector differential equation are given in this paper. A new design methodology for rotation vector-based attitude updating algorithms is proposed. Different vibratory dynamics and high rate manoeuvre roller coaster experiments were conducted to validate the effectiveness of the new algorithm. The results demonstrate the high accuracy of the new algorithm compared with conventional coning correction methods. The proposed algorithm can also be used in high accuracy attitude computation of a post-processing system, especially when the output frequency of the gyro is limited.