Professor Jia-Richards Leads Autonomous SmallSat Project Funded by NASA
Michigan Aerospace Engineering Assistant Professor Collaborates with NASA’s Jet Propulsion Laboratory As Part of NASA’s University SmallSat Technology Partnerships Initiative
University of Michigan Aerospace Engineering Assistant Professor Oliver Jia-Richards has begun collaborating with NASA’s Jet Propulsion Laboratory (JPL) in Southern California through a University SmallSat Technology Partnerships (USTP) award, supported by NASA’s Small Spacecraft Technology (SST) program. Over the course of two years, the University of Michigan researcher will receive up to $450,000 in grants to work alongside JPL’s Keenan Albee and David Sternberg.
The project, “SmallSat Steward: Autonomous Online Learning and Planning for Safe Inspection of Cislunar Stations,” will be researching and adapting the use of algorithms for small spacecraft operating in proximity to space stations like the International Space Station, future commercial low-Earth orbit destinations, or potential stations in lunar orbit. The primary research focus lies in devising effective methods for inspecting the exterior of a space station while accounting for potential thruster degradation or other onboard failures.
Professor Jia-Richards raised a critical question in this context: How can we conduct inspections of the exterior of the station without endangering human lives or incurring exorbitant costs?
Traditional methods, such as spacewalks or using robotic arms attached to the station with limited movement and visibility, present their own challenges.
To address these issues, Jia-Richards is exploring the use of a free-flying small spacecraft that can act autonomously. This spacecraft would navigate around the outside of the concept space station, conducting inspections without jeopardizing astronaut or station safety and potentially at a much lower cost. The question then becomes what happens if something goes wrong while doing this inspection? Specifically, the researchers are looking at the possibility of unexpected thruster failure or degradation.
“All the trajectories that we have planned for this free-flying spacecraft are typically designed under the assumption about a particular level of performance that we are going to achieve from the propulsion system,” Jia-Richards explained. “So, if the propulsion system degrades or fails, then those trajectories are no longer valid, and the spacecraft won’t follow them anymore.”
With the objective of exploring the use of reinforcement learning techniques, the team plans to have the small spacecraft operate autonomously when changes occur. Jia-Richards and JPL collaborators will also be using flight-like and emerging onboard computers in order to demonstrate the method on the expected computational capabilities of future free-flying small spacecraft. This will allow it to react and respond to various changes such as thruster degradation or failure. In the event of thruster failure, the spacecraft would still try to carry out the mission, recognize an issue, and use that information to update its internal model of how it expects to act. It then would adjust its trajectory or control in an attempt to either continue the mission with minor modifications or, in the case of a significant problem, safely eject itself away from the station to prevent damage.
“The goal is for us to do the theoretical algorithm development here at U-M, and perhaps some of the initial testing as to how we might implement it onto CubeSat-like hardware or small satellite-like hardware. Then we would do hardware testing at JPL to demonstrate the algorithm working on flight-like hardware,” Jia-Richards explains.
Looking ahead, this project is envisioned to conduct “hardware-in-the-loop testing” using JPL’s Small Satellite Dynamics Testbed to demonstrate algorithm performance on constrained processors. The desire of the project is to eventually do in-space testing on a flight demo as a continuation of the project. Jia-Richards expanded on this to explain the number of different paths that might take. One may be developing a CubeSat in house to test it or collaborating with another company to build it. Another avenue involves the use of NASA’s Astrobee, small robots aboard the International Space Station capable of testing software and hardware in space.
An immediate follow-up could be to have these robots test out this program on the International Space Station to demonstrate them in the space environment. Eventually, however, the team would like to demonstrate their autonomous algorithm with an external free-flying spacecraft. This forward-looking approach aims to validate the technology’s effectiveness in real-world space conditions.