Comparison of multiobjective optimization algorithms for mobility behaviors in autonomous robot systems

Innovative mobility algorithms for autonomous robots have been developed to address civil applications such as disaster relief in the past. Using sophisticated development methodologies such as combinations of model-based as well as Software- and Hardware-in-the-Loop simulations help to reduce the gap between simulations and real world scenarios. An open issue regarding the mobility is to find the optimal parametrization considering multiple optimization goals. In this research work, we introduce and analyze the Mobility Evaluation and Parameter Optimizer (MobEPO). Multiobjective optimization algorithms are used to find optimal parameter sets. We present the approach and a proof-of-concept evaluation in a common exploration scenario. We compare three suitable optimization algorithms and describe their use for prospective steering algorithm design.