Autonomous thruster failure recovery on underactuated spacecraft using model predictive control

Abstract

Thruster failures historically account for a large percentage of failures that have occurred on orbit. These failures are typically handled through redundancy, however, with the push to using smaller, less expensive satellites in clusters or formations there is a need to perform thruster failure recovery without additional hardware. This means that a thruster failure may cause the spacecraft to become underactuated, requiring more advanced control techniques. A model of a thruster-controlled spacecraft is developed and analyzed with a nonlinear controllability test, highlighting several challenges including coupling, nonlinearities, severe control input saturation, and nonholonomicity. Model Predictive Control (MPC) is proposed as a control technique to solve these challenges. However, the real-time, online implementation of MPC brings about many issues. A method of performing MPC online is described, implemented and tested in simulation as well as in hardware on the Synchronized Position-Hold, Engage, Reorient Experimental Satellites (SPHERES) testbed at the Massachusetts Institute of Technology (MIT) and on the International Space Station (ISS). These results show that MPC provided improved performance over a simple path planning technique.