PhD Candidate · Aerospace Engineering · Texas A&M University
I build the guidance, navigation, and control that lets spacecraft fly precisely — from cislunar trajectory optimization to high-precision state estimation, validated from theory all the way to hardware.
I'm a PhD candidate in Aerospace Engineering at Texas A&M University (GPA 4.0/4.0), working at the intersection of control theory, state estimation, and trajectory optimization. My research focuses on making spacecraft maneuvers fast, accurate, and robust to uncertainty.
Much of my work centers on cislunar trajectory optimization and model-predictive control, high-precision numerical differentiation for dynamical systems, and estimation techniques that hold up in real, noisy environments. I care deeply about closing the gap between elegant theory and flight-ready software — including hardware-in-the-loop validation on real embedded platforms. These same guidance, navigation, and control methods carry over naturally beyond spacecraft — to drones, automotive, and robotics systems — and those are areas I'm especially excited to work in.
Before Texas A&M, I earned my BS in Mechanical Engineering from Waseda University in Tokyo, and I've since worked on navigation systems at VectorNav and spacecraft landing dynamics at JAXA/ISAS.
My work spans cislunar trajectory optimization, optimal control under uncertainty, high-precision differentiation, and state estimation. Below are selected projects across those areas. Detailed write-ups and slides coming soon.
High-order complex-step (CX) state-transition tensors that propagate orbital uncertainty under J2 and atmospheric drag — built without a closed-form flow map. Benchmarked across five coordinate sets (Cartesian, COE, MEE, GEqOE, MRP-MEE) against Monte Carlo, matching its accuracy at 1–2 orders of magnitude lower cost.
Nonlinear model-predictive control (NMPC) that corrects cislunar transfers in the CR3BP under impulsive-maneuver uncertainty, blending pre-planned impulsive burns with low-thrust NMPC corrections. On a southern L2 halo-to-halo transfer it holds sub-kilometer tracking error at ~315 m/s total Δv, with IPOPT running up to 2× faster than MATLAB's fmincon.
A systematic comparison of two NLP solvers (MATLAB's fmincon and IPOPT) and three derivative methods — finite-difference, complex-step (CX), and analytical — on a minimum-Δv multi-impulse transfer between two Distant Retrograde Orbits in the CR3BP. The CX method nearly matches analytical derivatives while being far easier to implement, and IPOPT converges faster than fmincon with accurate derivatives.
An extension of the complex-step method for partial derivatives of smooth vector functions of n variables, with novel multivariable second- and higher-order formulas. In 64-bit arithmetic it reaches ~15-digit accuracy for first derivatives, 10–13 for second, and 5–8 for third — with no operator overloading (MATLAB/Python/C++), far cheaper than automatic differentiation and more accurate than finite differences.
A multiple-shooting, multi-impulse framework for minimum-Δv cislunar trajectory optimization, combining impulse regularization, complex-step sensitivities, and Jacobian coloring. By exploiting the constraint Jacobian's sparsity, coloring cuts per-iteration function evaluations (52 → 14 for a five-node transfer) for a ~2× speed-up, generating families of locally-optimal transfers between southern L2 halo orbits — the best exploiting a lunar gravity assist.
A gyro-free method for determining a spacecraft's angular velocity and acceleration from an array of triaxial accelerometers, using linear least squares to estimate functions quadratic in angular velocity. Establishes minimum sensor requirements, a way to resolve the spin-axis sign, and a linearized covariance approximation that matches Monte-Carlo results.
Post-Kalman smoothing algorithms that doubled state-estimation accuracy for real-time navigation, plus high-precision geoid-undulation modeling with the EGM2008 gravity model. (VectorNav)
Physics-based mass-spring-damper models of propellant sloshing during lunar landing, with stability analysis and landing-gear control laws to prevent overturn. (JAXA/ISAS)
Alignment-monitoring payload designed and prototyped for stratospheric flight, with structural analysis validating survival in near-space conditions. (Waseda)
Research and engineering roles across academia and industry, in the US and Japan.
Texas A&M University · College Station, TX
VectorNav Technologies · Dallas, TX
JAXA / ISAS · Kanagawa, Japan
Waseda University · Tokyo, Japan
Peer-reviewed journal and conference papers. Google Scholar ↗
Journal of Guidance, Control, and Dynamics (JGCD). Under review
2026 American Control Conference (ACC), IEEE, 2026.
Aerospace Science and Technology, Vol. 180, Art. no. 113985, 2027.
The Journal of the Astronautical Sciences, 73(3), 2026.
2025 American Control Conference (ACC), IEEE, July 2025, pp. 3807–3812.
AAS/AIAA Astrodynamics Specialist Conference, Broomfield, CO, Aug. 2024.
AAS/AIAA Astrodynamics Specialist Conference, Big Sky, MT, Aug. 2023.
Texas A&M University · 2022 – Present · GPA 4.0/4.0
Random Dynamical Systems · Space Domain Awareness · Celestial Mechanics · Spacecraft Dynamics · Optimal Attitude & Orbital Maneuvers · Estimation of Dynamic Systems
Waseda University, Tokyo · 2018 – 2022 · GPA 3.6/4.0
Mechatronics · Fluid Dynamics · Satellite Design · C Programming
Interested in collaborating on spacecraft GNC, or in my work for a role? I'd love to hear from you.
or reach me directly at koyayamamoto123@gmail.com