Integrated Low-Resource Estimation-Control Architecture for Attitude Regulation of CubeSats

Small satellites such as CubeSats are increasingly being used for Earth observation, communication, and deep-space missions, but their limited onboard computational resources pose challenges for traditional attitude estimation and control architectures. This work proposes an integrated low-resource guidance, navigation, and control framework that combines a reduced-order Kalman filter (ROKF) with a discrete-time linear quadratic regulator (DLQR) for real-time attitude regulation. The estimator exploits the structure of linearized rigid-body dynamics to separate a fast six-state loop on attitude and angular rate from a slow three-state gyro-bias loop, significantly reducing covariance dimension and update cost relative to conventional multiplicative extended Kalman filters. The same linear model is reused for both estimation and control, enabling code and memory sharing and simplifying implementation on microcontrollers with limited RAM, flash, and floating-point performance. Software-in-the-loop simulations with realistic sensor noise, eclipse-driven sun-sensor outages, and constant gyro bias demonstrate that the proposed architecture maintains steady-state pointing accuracy better than 0.2◦ root-mean-square while using modest control torques within CubeSat reaction-wheel capability. The results indicate that resource-aware co-design of estimator and controller can provide sub-degree performance without resorting to computationally intensive nonlinear filters, making the approach attractive for low-cost, resource-constrained space missions.