A generalized Bellman-Ford Algorithm for Application in Symbolic Optimal Control

Symbolic controller synthesis is a fully-automated and correct-by-design synthesis scheme whose limitations are its immense memory and runtime requirements. A current trend to compensate for this downside is to develop techniques for parallel execution of the scheme both in mathematical foundation and in software implementation. In this paper we present a generalized Bellman-Ford algorithm to be used in the so-called symbolic optimal control, which is an extension of the aforementioned synthesis scheme. Compared to the widely used Dijkstra algorithm our algorithm has two advantages. It allows for cost functions taking arbitrary (e.g. negative) values and for parallel execution with the ability for trading processing speed for memory consumption. We motivate the usefulness of negative cost values on a scenario of aerial firefighting with unmanned aerial vehicles. In addition, this four-dimensional numerical example, which is rich in detail, demonstrates the great performance of our algorithm.