Constrained Spacecraft Attitude Control Using Sequential Convex Programming

Abstract

This thesis describes how sequential convex programming can be used to guide and controlthe attitude of a spacecraft which is subject to pointing constraints. The constraints are expressed as convex contraints, and the dynamics are linearized and discretized using direct collocation and a piecewise linear approximation of the control to allow the problem to be solved quickly using sequential convex programming. The sequential convex program is initialized using the analytical solution to a simplified problem so that convergence is achieved in fewer subproblem iterations compared to other initialization methods. In addition, the collocation points are redistributed at each iteration to increase the density of collocation points near active constraints, preventing constraint violation. Finally, a simulation setup that includes disturbance torques and sensor and actuator dynamics is used demonstrate that a satellite can be controlled reliably using sequential convex programming without ever violating a constraint in a real-world environment.