Quantification and Minimization of the Simulation-Reality-Gap on a BRIO Labyrinth Game

Constantin Bergatt, Jan Hendrik Metzen, Elsa Andrea Kirchner, Frank Kirchner

In: Proceedings of the first International Workshop on Learning and Data Mining for Robotics. International Workshop on Learning and Data Mining for Robotics (LEMIR-09) 1st September 7-11 Bled Slovenia 9/2009.


In this paper we present a new method, the so-called e-tunnel, that can be used to quantify the simulation-reality-gap by comparing the behavior (the state-trajectory) of a baseline system (for instance a real robotic system) and a model of this system (e.g. a physicbased simulation). With this e-tunnel, the impact of a change of the model's parameter can be analyzed. Furthermore, we present an approach to automate the optimization of these parameters and present some results obtained on a robot system that is based on a BRIO labyrinth game.

Weitere Links

090831_Quantification_and_Minimization_of_the_Simulation-Reality-Gap_on_a_BRIOR__Labyrinth_Game_LEMIR_Bergatt.pdf (pdf, 611 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence