A toolbox for multi-objective planning in non-deterministic environments with simulation validation

Mobile robots are often subject to several costs and tasked with multiple objectives with different priorities when operating in realistic situations. We recently proposed and theoretically analyzed a planner handling multiple costs and multiple tasks expressed as syntactically co-safe linear temporal logic formulas. In this paper, we present an easy-to-use toolbox which is capable of solving multi-objective planning in non-deterministic environments where multiple tasks have to be accomplished. We experimentally evaluate this planner in a simulation scenario, and we moreover analyze its robustness to modeling errors in the underlying motion models. We show how this theoretical framework can be instantiated in a practical scenario and demonstrate its robustness.