1: \begin{abstract}

2: This paper proposes two novel nonlinear attitude filters evolved directly

3: on the Special Orthogonal Group $\mathbb{SO}\left(3\right)$ able

4: to ensure prescribed measures of transient and steady-state performance.

5: The tracking performance of the normalized Euclidean distance of attitude

6: error is trapped to initially start within a large set and converge

7: systematically and asymptotically to the origin from almost any initial

8: condition. The convergence rate is guaranteed to be less than the

9: prescribed value and the steady-state error does not exceed a predefined

10: small value.%

11: \begin{comment}

12: The constrained error is relaxed and converted to a new form, called

13: transformed error, which is unconstrained.

14: \end{comment}

15: {} The first filter uses a set of vectorial measurements with the need

16: for attitude reconstruction. The second filter does not require attitude

17: reconstruction and instead uses only a rate gyroscope measurement

18: and two or more vectorial measurements. These filters provide good

19: attitude estimates with superior convergence properties and can be

20: applied to measurements obtained from low cost inertial measurement

21: units (IMUs). Simulation results illustrate the robustness and effectiveness

22: of the proposed attitude filters with guaranteed performance considering

23: high level of uncertainty in angular velocity along with body-frame

24: vector measurements.

25: \end{abstract}