Papers
5
Total Citations
219
H-Index
4
About
Tae Ha Park is a leading researcher in autonomous spaceborne navigation, specializing in machine learning for spacecraft pose estimation—a critical technology for future on-orbit servicing, debris removal, and space logistics. His most impactful contribution is the development of the SPEED+ dataset (141 citations), a next-generation benchmark that bridges the domain gap between synthetic training data and real space imagery, enabling robust vision-based navigation. Park pioneered the integration of neural networks with adaptive unscented Kalman filters, creating hybrid systems that achieve reliable pose tracking of noncooperative, tumbling spacecraft—even under challenging lighting and motion conditions. His work on SPNv3 (Spacecraft Pose Network v3) further advances flight-ready, computationally efficient models that maintain accuracy across unseen spaceborne images. A key enabler of this research is the TRON robotic testbed at Stanford University, which Park helped develop to validate machine learning algorithms for optical navigation in realistic, hardware-in-the-loop scenarios. With over 200 citations across his core publications, Park’s contributions are shaping the next generation of autonomous spacecraft operations, making him a pivotal figure in the intersection of computer vision, robotics, and spaceflight engineering.
Research Focus
Key Achievements
Top Papers
- 1SPEED+: Next-Generation Dataset for Spacecraft Pose Estimation across Domain Gap141 citations · 2021
- 2
- 3
- 4
- 5Bridging the Domain Gap for Flight-Ready Spaceborne Vision3 citations · 2025