A new method makes robust control solvable when uncertainty changes with your decisions.

Engineers often need to control systems like satellites safely, even when conditions are uncertain. A major challenge happens when the uncertainty itself changes depending on the system's state or the decisions made. Math problems of this type are called generalized semi-infinite programs. Existing ways to solve them either make narrow assumptions or become too slow. To fix this, researchers developed a new framework. It smooths tricky, divided conditions into simpler, differentiable existence constraints over a fixed set. This lets standard, off-the-shelf software solve the problem using adaptive discretizationadaptive discretizationA method that breaks down complex problems into simpler parts over time to find solutions, often used in iterative optimization., or cutting-plane methods. By treating the system's path as part of the uncertainty, it handles state-dependent situations. The team tested it successfully on a benchmark problem and a satellite de-tumbling task with changing inertia.

Why it matters

Engineers designing control systems for things like satellites face uncertainty that shifts based on the system's state. Current tools often fail here, but this work offers a broader approach that could make such systems more reliable and easier to optimize.

Questions answered