HARISHVERNEKAR

Developed a 3-layer GNC architecture (ARISE) for CubeSat swarm separation from an ESPA ring. Layer-1 uses a Port–Hamiltonian SE(3) damping controller with energy-tank passivity guarantees to rapidly quench post-deployment tumble rates (4°/s → 0.05°/s) and drift velocities (0.10 m/s → 2 mm/s) within 2 minutes, enabling clean handoff to the attitude control layer — even under actuator delays and saturation.

Designed a complete mission architecture to deploy 8 CubeSats from a Powered ESPA ring as a GEO secondary payload, intercept asteroid (99942) Apophis before its April 2029 Earth flyby at just 169 m/s ΔV, and capture stereoscopic surface imagery at 0.15m resolution (closing velocity 8.76 km/s). The swarm uses ROE-based formation control with genetic-algorithm-optimized positioning for 100% stereo coverage. Post-flyby, the deputies autonomously rendezvous and redock on the ESPA platform for reuse.

Extended the classic 6-element ROE state with inter-satellite baseline shape descriptors to enable fuel-optimal swarm reconfiguration. Each deputy's orbit is controlled using impulsive maneuvers mapped through state-transition matrices, with total ΔV minimized via numerical optimization. The framework achieves ±0.002% position/velocity conversion accuracy and enables real-time formation shape control for multi-angle imaging scenarios.

Applied deep reinforcement learning to solve the multi-deputy rendezvous problem post-Apophis flyby. The RL agent autonomously plans collision-free, low-ΔV trajectories for 8 CubeSats to achieve 20m proximity with the ESPA platform — replacing hand-tuned guidance laws with a learned policy that adapts to uncertainties and unmodeled accelerations in real time.

Designed a cislunar satellite formation using the CR3BP framework with an [E, L1, M, E] mission itinerary. Deputies are placed on stable eigenvectors from the monodromy matrix of a Lyapunov reference orbit. Formation station-keeping uses LQR with PSO-optimized gain matrices, maintaining bounded triangular geometry with minimal continuous thrust while leveraging natural manifold dynamics for fuel-efficient transfers.