A Complementary Filter Based on Multi-Sample Rotation Vector for Attitude Estimation

Abstract

Attitude estimation is an important issue in motion control and inertial navigation for spacecrafts, aircrafts, self-driving vehicles, underwater robots, and portable navigation systems. Many researchers have proposed various attitude estimation approaches, but rarely considered their applicability to high or ultra-high rotation scenarios. Conventional attitude estimation in a strap-down inertial navigation system is based on the integral calculation of rotation kinematics, which unavoidably suffers from accumulative and noncommutativity errors in high dynamics and high rotations. In this paper, an improved algorithm of multi-sample equivalent rotation vector using angular rates instead of angular increments is proposed to calculate attitude angles for enhancing accuracy. An advanced complementary filter using the angular rate-based rotation vector is further developed to implement sensor fusion for attitude determination under high or ultra-high rotations. The effectiveness and advantages of the proposed attitude estimation methodology are validated through simulation experiments.