Special Seminar with Dr Kaushik Prabhu

Formation flying is a critical technology for future space missions. Distributed estimation will play a key role in achieving multi-spacecraft mission objectives. This work develops two types of estimation algorithms: a distributed batch filter for static parameter estimation and a distributed real-time filter for dynamic state estimation. 

For batch filtering, we begin by developing the Least Absolute Deviations (LAD) estimator that serves as a robust alternative to the well-known least squares technique. A sampling distribution theory for the LAD estimates is also introduced. For applications in multi-agent systems, the Distributed (D-) LAD estimator is derived. In the D-LAD algorithm, individual agents utilize local measurement data and exchange information with their immediate neighbors to collaboratively compute the LAD estimate.

For distributed real-time filtering, we introduce the Distributed Absolute and Relative Estimation (DARE) algorithm for autonomous inertial localization of spacecraft formations. DARE enables full formation state estimation at each spacecraft in the presence of local observability and communication constraints. Further, a computationally efficient version called the Sparse (S-) DARE algorithm is also derived for implementation on nanosatellites where resources are limited.