Singularity avoidance in spacecraft with single gimbal CMGs using RRT-Based steering laws

A cluster of single-gimbal control moment gyroscopes (CMGs) is commonly employed on spacecraft for generating large torques. It is well-known that the presence of geometric singular configurations presents challenges when using CMGs for high-precision maneuvers. In this paper, we investigate the use of a Rapidly-Exploring Random Trees (RRT) algorithm for singularity avoidance. The analogy exploited in this approach is that between a CMG cluster singularity and a solid obstacle in a typical motion planning problem. We demonstrate singularity avoidance, and subsequent improved tracking of commanded torques, using numerical simulation. The performance of RRT is compared with conventional singularity avoidance and escape steering laws. A special case of starting the CMG cluster from a singular gimbal angle condition is also presented in this study.