Team B0: Celestial Positioning
Carnegie Mellon ECE Capstone, Spring 2026 | Julius Arolovich, Seungyeon Lee, Eddie Zhang
This week I focused on the final presentation, building out the slides, prepping my delivery since I’m the presenter, and double-checking every number in the deck against our raw data to make sure nothing is off.
We’re on schedule. All results are under threshold: 2.1 km median position error (target 10 km), 0.32 px reprojection error, 500 μs time sync (target 50 ms), 4.5 W power draw (target 10 W). Only two things left: the moored balloon experiment and demo day!
This week, I upgraded the celestial positioning pipeline with ASTAP as the primary plate solver (1–2s vs 10–60s for astrometry.net) with automatic fallback when ASTAP can’t solve. I replaced attitude extraction with direct rotation matrix extraction. I alidated end-to-end by plate-solving a real iPhone star photo and a synthetic FITS image from NASA.
Next week I need to test the full pipeline on the Pi with the existing camera calibration and live IMU/mag data to get an actual lat/lon fix with ground truth comparison.
I am on track. The core math is all implemented and tested, camera calibration is done, and sensor data is validated. The remaining work is Pi integration and accuracy evaluation.
This week, I focused on improving the robustness of the attitude estimation pipeline and validating its consistency across different image inputs. I refined the integration between the plate solver output and our calibrated camera model, which helped reduce inconsistencies in the computed orientations.
I need to begin connecting the attitude outputs to the next stage of the pipeline and figure out how to convert these orientations into preliminary latitude and longitude estimates. This is is critical for meeting our system accuracy goals.
We verified our subsystem by testing star detection robustness under noise, blur, and occlusion, and by validating plate solving and attitude estimation using known WCS outputs, measuring angular residual errors and solve success rates. For validation, we use ground-truth datasets from NASA SkyView and Stellarium to compare estimated boresight RA/Dec against known values and compute angular and position error relative to the ≤10 km requirement.
This week, I primarily focused on implementing and testing the plate solving functions to accurately extract the camera’s attitude. After successfully completing camera calibration, I began integrating the calibration parameters into the pipeline and testing how they affect the computed camera attitudes. I also started evaluating the ASTAP plate solver that Prof. Mukherjee recommended during our meeting last week, with the goal of improving the efficiency and runtime of our current solving pipeline.
In the coming week, I plan to make the attitude estimation functions more robust so that they can consistently produce reliable camera orientation estimates. Once the attitude estimation is stable, we will be able to move forward to the next stage of the pipeline, which involves estimating global longitude and latitude from the celestial observations.
This week I focused on the visual processing portion of our celestial positioning system. I worked on refining the visual pipeline architecture for our report, including the image capture, preprocessing, star detection, plate solving, WCS boresight extraction.
My progress is currently on schedule. The work I completed this week aligns with my role in image processing and vision, and it gives us a much clearer structure for implementing and validating the visual pipeline.
This week, I investigated optimal camera configurations for reliable star detection under nighttime conditions. I researched how exposure time, gain, and aperture impact signal-to-noise ratio, motion blur, and star visibility, especially given the UAV’s motion. I explored trade-offs between longer exposure, which gives brighter stars but increased motion blur, and shorter exposure, which has reduced blur but lower signal strength.
I am currently on schedule, as this week was intended to narrow down hardware and imaging parameter considerations before implementation. In the immediate next steps, I plan to begin testing different exposure and gain combinations on sample nighttime imagery and evaluate detection performance quantitatively.
This week, I focused on building and validating the core algorithmic pipeline for our celestial positioning system. I implemented star detection using preprocessing, adaptive thresholding, connected components, and centroid extraction on a stock starfield image. I then plate-solved the image using astrometry.net and integrated the resulting WCS file into our Python pipeline. The system achieved a mean angular error of approximately 0.0014 degrees, confirming that our detection and pose estimation steps are functioning accurately.
My progress is on schedule, as this week’s goal was to establish a working baseline registration and attitude estimation pipeline before implementing our own star identification method. Next week, I plan to replace the WCS-based labeling with a triangle-based star matching algorithm and validate that we can achieve similar accuracy independently.
This week, I focused on completing and refining the use case and user requirements for our design proposal. As part of this process, I explored the technical implications of our requirements, particularly calibration accuracy, environmental assumptions, and realistic localization error bounds. I also began foundational work for my role by doing initial research on star detection and tracking algorithms suitable for low-light conditions.
I am currently on schedule, as this week was intended for proposal refinement and background research. In the immediate next steps, I plan to prototype simple star detection and preprocessing pipelines using sample nighttime imagery. I also plan to further explore camera calibration workflows and imaging parameter tuning to prepare for early system-level testing.