Toolbox Description

The Automated Reachset Optimal Control (AROC) toolbox automatically constructs formally verified controllers for solving reach-avoid problems using reachability analysis. The toolbox considers general nonlinear systems with disturbances and input constraints. AROC is implemented in MATLAB, and the CORA toolbox is used to calculate reachable sets. The following points summarize the main capabilities of the AROC toolbox:

  • Motion Primitive Based Control

    For motion prmitive based control optimial controllers for many motion primitives are synthesized offline and then applied for online control using a maneuver automaton. Currently, the three motion primitive based control algorithms optimization based control, convex interpolation control, and generator space control are implemented.

  • Model Predictive Control

    For model predictive control the controller is constructed and verified at runtime, where the computation time required for the optimization based construction of the controller as well as the verification with reachability analysis is explicitely considered by the algorithm.

  • Motion Planning

    AROC contains an implementation of a maneuver automaton, which can be applied for online control using the motion primitives constructed offline. For autonomous road and marine traffic applications the toolbox provides interfaces to CommonRoad and CommonOcean, making it easy to test the algorithms on different traffic scenarios.

  • Safe Terminal Regions

    Also algorithms for constructing controllers that stabilize a system around a given equilibrium point and aim to maximize the size of the corresponding safe terminal region are part of AROC. The resulting terminal regions are well suited to be used in combination with model predictive control approaches.

  • Conformant Synthesis

    To guarantee that the synthesized controllers are safe for the real system and not just for a given model we require over-approximative models that enclose all possible behaviors of the real system. The conformant synthesis algorithms in AROC can be used to construct such over-approximative models automatically from measurements of the real system.

PLACEHOLDER

Videos

    Conference Presentation (HSCC 2021)

    AROC in Action

PLACEHOLDER

Publications

[1] N. Kochdumper, F. Gruber, B. Schürmann, V. Gaßmann, M. Klischat, and M. Althoff. AROC: A Toolbox for Automated Reachset Optimal Controller Synthesis In Proc. of the International Conference on Hybrid Systems: Computation and Control, 2021. [ .bib | .pdf ]