In this work we consider uncertain optimization problems where no probability distribution is known. We introduce the approaches RecFeas and RecOpt to such a robust optimization problem, using a location theoretic point of view, and discuss both theoretical and algorithmic aspects. We then consider both continuous and discrete problem applications of robust optimization: Linear programs from the Netlib benchmark set, and the aperiodic timetabling problem on the continuous side; intermodal load planning, Steiner trees, periodic timetabling, and timetable information on the discrete side. Finally, we present the software library ROPI as a framework for robust optimization with support for most established mixed-integer programming solvers.