Quaternion based LQR for Free-Floating Robots without Gravity

Shubham Vyas, Bilal Wehbe, Shivesh Kumar

In: 6th CEAS Conference on Guidance, Navigation, and Control. CEAS Conference on Guidance, Navigation and Control (EuroGNC-2022) May 3-5 Berlin Germany CEAS 5/2022.


Quaternions are commonly used for rotation representation as they avoid the singularities found in the Euler angles representation and are more compact than using rotation matrices (for storage, operations, and constraints required). However, it is difficult to use quaternions in linear control approaches due to the inherent unit length constraint of the representation. Quaternion-based linear control has been previously used for single rigid body control such as quadrotors and satellite attitude control. In this paper, we provide an analytical method for linearizing multibody freefloating robotic systems without gravity using a quaternion-based rotation representation for the floating base. This linearization is then used for deriving a Linear Quadratic Regulator (LQR) based controller. The LQR is optimal in the local neighbourhood of the linearization and is globally asymptotically stable for such systems. The utility of this method is demonstrated using two examples from different robotic domains: space and underwater robotics.


EuroGNC_2022_Quaternion_LQR_Paper_FinalVersion.pdf (pdf, 1 MB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence