Journal
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
Volume 66, Issue 10, Pages 4998-5005Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TAC.2020.3047553
Keywords
Algebraic/geometric methods; Kalman filtering; nonlinear observability; nonlinear systems; robotics
Ask authors/readers for more resources
This article discusses the observability properties and observer design for rigid body attitude under conditions of partial inertial sensing. The lack of first-order observability hinders standard observers from achieving global convergence, requiring a more suitable approach for observer design. The proposed approach is validated through numerical and experimental results, showing improvements in error convergence.
The aim of this article is to discuss the observability properties and observer design for the attitude of a rigid body, under conditions of partial inertial sensing. In particular, we introduce an observability analysis tool for the attitude dynamics when only accelerometer and gyroscope measurements are available, as in several robotics applications. In various scenarios, in fact, the measurement of the magnetic field via a magnetometer is unreliable, due to magnetic interferences. Herein, we first focus on a formal observability analysis, which reveals that the target dynamics is weakly locally observable, but not first-order observable. The lack of first-order observability prevents standard observers from achieving global convergence. Therefore, we discuss a more suitable approach for observer design to deal with this problem. The proposed approach is validated by providing numerical and experimental results. The former show that the proposed approach is able to achieve convergence (final error 0.004%). Experiments validate our inference about observability and show the improvements brought by the proposed approach concerning the error convergence (final error 0.15%).
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available