Develped MATLAB code to extract the trajectories of CubeSats using deployment footage from any uncalibrated camera. Authored and presented on this research at an American Astronautical Society Conference and won second place in Student innovations in GNC.
Three 1U CubeSats and one 3U CubeSat can be seen over Earth in this frame from the deployment video.
I generated a mean filter by combining every frame of the video, then subtracted that from each individual frame. In deployment videos the CubeSats are moving much faster than the Earth rotating beneath them, so the CubeSats hardly appear in the filter.
Because the CubeSats don't appear in the filter, they don't get subtracted from each frame like the background. That means that the brightest pixels remaining after the filter is subtracted are the CubeSats.
CubeSats detected in every frame where they have a red box around them. Algorithm struggles in the middle because I did not use adaptive thresholding, I simply used a constant value to threshold because there were so many clouds in this video.
Found the average distance between opposite corners on the cube and correlated that to the number of pixels making up the object in the image.
Found the line of best fit through all of the datapoints for the best estimate of PPM in each frame.
Three of the CubeSats are predicted to be very nearly the same depth which is accurate since they were deployed in a line. One of the 1U CubeSats is calculated to be farther than the others since its face is perpendicular to the camera which lowers the accuracy of the length estimate.
GUI takes in the 2D image coordinates of the corners then finds the direction cosine matrix (DCM) which represents the cube's pose. CubeSats follow known dimensions, and by using the planar components of the rotated cube I can precisely solve for PPM and depth.
Authored and presented on a paper titled Extracting Cubesat Relative Motion Using In Situ Deployment Imagery at an American Astronautical Society Guidance Navigation and Control Conference in Breckenridge, CO. On the left is my mentor and (former) PhD candidate Dr. John Gaebler and on the right is my advisor and renowned professor Dr. Penina Axelrad.
