AROC: a toolbox for automated reachset optimal controller synthesis

Abstract

We present a MATLAB toolbox for Automated Reachset Optimal Control (AROC) that automatically synthesizes verified controllers for solving reach-avoid problems using reachability analysis. The toolbox implements two different types of control approaches: When using our verified model predictive controller, a feasible control law is constructed and verified on-the-fly during online application of the system. For motion-primitive-based control, on the other hand, controllers for many motion primitives are synthesized offline and then used for online motion planning with a maneuver automaton. Since our toolbox considers general nonlinear systems with input constraints, state constraints, and bounded disturbances, it is applicable to a very broad class of systems, as we demonstrate with several numerical examples. AROC is available at https://aroc.in.tum.de.