Experience & Education · Tae Ha Park

Experience

Developing an in‑house CubeSat AOCS system and its software & hardware validation environments
Developing an intelligent system for autonomous vision‑based RPO about unknown resident space objects

Developed robust deep learning models and GN&C algorithms for vision-based rendezvous with known and unknown targets in space
Developed and calibrated the Testbed for Rendezvous and Optical Navigation (TRON) facility at SLAB that is capable of physically simulating spacecraft proximity operations with a mockup satellite mdodel under high-fidelity spaceborne illumination settings and estimating high-accuracy pose labels
Developed the next-generation open-source datasets (e.g., SPEED+, SHIRT) using TRON to train and validate spaceborne vision-based deep learning and GNC algorithms with emphasis on robustness across domain gap
Organized the second international Satellite Pose Estimation Competition (SPEC2021) in collaboration with the European Space Agency

Constructed a satellite rendezvous simulator and scene renderer based on Unreal Engine 5 to create images for training and validating neural networks for spaceborne computer vision

Developed a method to assess the functionality of steel anchors embedded within a concrete dam based on the Performance-Based Testing using spectral analysis, spectrogram, and model verification
Analyzed the earthquake response of Monticello dam by constructing and evaluating a lumped element model of dam, reservoir and a spillway

Developed a method to accelerate the modal convergence of the eigen-characteristics of uniform and non-uniform rods carrying various lumped attachments

Thesis: Robust machine learning for vision-based navigation about non-cooperative resident space objects

Conducted research on deep learning, computer vision, spacecraft swarm navigation and optimization.