Rutgers professor wins award from NASA
Xiaoli Bai, an assistant professor in the Department of Mechanical and Aerospace Engineering at Rutgers University, was recently named as the recipient of the NASA Early Career Faculty Award.
The NASA Early Career Faculty Award is awarded to untenured, assistant professors studying early-stage space technologies. Bai’s research was 1 of 9 proposals that were selected across the country.
Her research proposal, titled “A Holistic Bayesian Framework for Intelligent Calibration of Constellations of Sensors,” which falls under NASA’s category of Intelligent Calibration of Sensor Constellations, focuses on technology that would enable autonomous sensor calibration, providing “richer information at higher resolutions than using an individual sensor on one spacecraft," according to NASA's site.
While sensors are calibrated before satellites are launched into orbit, vibrations during the launch can disrupt the calibration performed on the sensitive instruments, she said.
“So the special situation for us is, during the launch, the vibrations can change many parameters,” Bai said. “So we basically want to come up with some way you can do that in orbit automatically."
The emphasis in her study is the human-free aspect. That is, the autonomous nature of the calibration, she said.
“We want to avoid on-ground operations, which are what we currently do,” Bai said, pointing to an ongoing problem with today’s systems. “Should I observe that spot, some flood or some fires that are going on, or should I do some of my calibration maneuvers? So there’s some trade-off. Our idea is to make this autonomous.”
Bai’s proposed study also improves on existing autonomous calibration techniques, she said.
“Our model is going to overcome some limitations that the current model uses,” Bai said. “The current models very often assume some parametric models. Like for cameras, the distortion is very often assumed to be a polynomial model. But, in fact, many cameras have a wider field of view and those models will have large errors. So our method, in fact, doesn’t need to make those assumptions.”
Bai also pointed out another advantage in her model compared to existing autonomous calibration methods.
“Why we call it ‘holistic' is because we emphasize the uncertainty quantification. I give you an estimation on, for example, the focal length, but I’ll also tell you how confident I am. So our framework, whenever we provide you an answer we will also provide you our certainty about that," she said.
It may be surprising to note that many existing models do not address these issues already. Bai said only “a few do, but maybe they do in only one part of the solution. Our way, we call it ‘holistic’: we start from every part and emphasize the uncertainty quantification.”
There are many immediate applications of the research, providing an example of observing a flood via satellite imagery.
“If you have one satellite, you can only see a part of (the flood). So what you can do is, you can configure (small satellites) as a formation so each provides some information about the target that you want to see, but if combined together, it can give you much better pictures,” she said.
Bai noted that future applications of the research are not limited to satellites in Earth orbit.
“I suppose you want to watch something beyond the solar system, and also in the solar system,” Bai said. “Our idea is, we can have many cheap satellites, they work as a group, they talk to each other and they watch (the target) for us.”
This is not the first award that Bai has received for her research in autonomous control in aerospace engineering. She was also the winner of the Young Investigator Award in 2016 from the Air Force Office of Scientific Research (AFOSR) for her research on orbital prediction and satellite collision avoidance.
Bai’s latest research also aligns with her interests, which focuses on computationally efficient methods for dynamics and controls of aerospace systems, she said.
“We are bringing in new methods like machine learning, how you combine machine learning with the existing physics models so you can predict better and avoid future collisions,” she said.
As for the future, Bai said she hopes to improve her method and continue testing her model, whether that be through observing Earth’s environment or even the Sun’s solar activities.
Bai also provided advice for students studying aerospace engineering.
“Learn the fundamentals,” she said. “Nothing can replace what you derive, the very fundamentals. When I teach undergraduate classes here, I always emphasize to students many times to do whiteboard derivations.”
Comments powered by Disqus
Please note All comments are eligible for publication in The Daily Targum.