At a Glance

• Texas Spacecraft Laboratory collaboration with NASA Johnson Space Center (JSC) for follow on research to Seeker Vision. Research also funded by Dr. Maruthi Akella in the Controls Group For Distributed And Uncertain Systems.

• Seeker Vision was only capable of computing relative azimuth and elevation of a target spacecraft, not full 6 degree of freedom relative pose.

• Full relative pose will enable more complex autonomous proximity operations from free-flying spacecraft.

• Critically, the system must still be able to run onboard CubeSat-level hardware and is thus highly constrained computationally.

• First author on a paper covering the model training pipeline for this reseach, published at the 2021 IEEE Aerospace Conference in March 2021. Read it here.

• Personal Contributions:

  • Aided in design of an end-to-end model training pipeline, supported by AWS, to enable rapid, iterative model development in the cloud.
  • This pipeline consists of 4 main tools, all open source:
    • ravenML
    • jigsaw
    • bran
    • starfish
  • Led design and development of ravenML, a plugin based Python CLI for training machine learning models quickly and easily.
  • Served as team lead for one year, from January 2020 to January 2021