On active set algorithms for solving bound-constrained least squares control allocation problems

Control allocation problems, in which optimal actuator values are assigned based on desired control actions and actuator constraints, are typically formulated as constrained optimization problems. Initially, only approximate solutions to the problems were deemed tractable for real-time applications. However, active set algorithms have been identified as a means for solving control allocation problems in real-time applications. In this paper a modified active set algorithm for the solution of bound-constrained least-squares problems is presented, which has numerous advantages particularly for time-varying control allocation problems. A vehicle dynamics control system for rollover mitigation is used as an example.