Star Tracker: Next-Generation Single-Pixel Star Sensor
January 2015 - June 2015
Centre for Advanced Photonics and Electronics(CAPE), University of Cambridge, UK
Department of Precision Instruments, Tsinghua University, China
This is my final semester design project done in both University of Cambridge and Tsinghua University, in which a single-pixel star tracker based on compressive sensing was developed. My bachelor thesis was on this topic and won the best Undergraduate Thesis in the Department of Precision Instruments.
Motivate
-
High-accuracy attitude determination is essential to satellite control especially for deep-space missions. One of the most effective strategies is to record continuous images of the field of view (FOV) using a star sensor, and then determine its orientation by matching stars in the FOV with those in an onboard star catalog.
-
The key step of star tracking is imaging. Tranditional camera scans every pixel to find stars, yet most pixels are nothing but noise.
-
Combining compressive sensing algorithm and Digital Micromirror Device (DMD) can provide a solution to position stars without scaning every pixel of the sky.
Ideate
-
Compressive sensing addresses the inefficiency of acquiring every sample by directly acquiring a compressed signal representation without going through the intermediate stage of acquiring all samples.
-
By randomly selecting pixels to scan using TI Digital Micromirror Device (DMD), we could calculate where the stars are based on compressive sensing reconstruction algorithm.
Early sketch of ideas
Typical structure of star sensor
System design of the single-pixel star tracker
Basic principle of compressive sensing
Develop
-
Set up optical system and designed DMD control module.
-
Simulated imaging & reconstruction(Orthogonal Matching Pursuit) process in Matlab.
Experiment set up
Apply ramdom DMD array to select pixels
Grid and reconstruct star image
Reconstruction quality v.s. number of measurements
Star tracking result
Star pattern recognition principal
Test
-
Reduced the number of image scan by 60%.
-
Successfully reconstructed star image from signal pixel detector and got track of stars.
Simulation & Experiment
Orientation recognition rate
Deliver
-
Award: Best Undergraduate Thesis in the Department of Precision Instruments.
-
Presentation: Presented on Tsinghua Precision Engineering Symposium.
Skill Sets
Digital Image Processing
Optical Design
Embedded System Design
Matlab
Signal Processing