Abstract
In recent years, the application of artificial intelligence in production processes is becoming increasingly important. Initially focused on the quality management process, artificial intelligence has experienced more usage in production control processes and other complex dynamic systems. In addition to learning speed, the robustness of the resulting algorithm is an important quality criterion.
This paper describes the steps towards a reinforcement learning trained \(Q\)-network that operates an electromechanical pinball machine. The aim is that the algorithm generates inputs for the system such that the resulting playing time and the associated point gain is maximized and shows a superior performance compared to a human player. This paper presents the planned learning concept and contains exemplary measurements to determine transition probabilities. Based on these results a Monte Carlo simulation should be used to train the \(Q\)-network and thus be able to get an optimal initialization for the practical realization.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Internet Pinball Database. https://www.ipdb.org/machine.cgi?id=2539. Accessed January 2020
Jaderberg, M., et al.: Human-level performance in 3D multiplayer games with population-based reinforcement learning. Science 364(6443), 859–865 (2019)
Kober, J., Bagnell, A., Peters, J.: Reinforcement learning in robotics: a survey. Int. J. Robot. Res. 32(11), 1238–1274 (2013)
Kodratoff, Y.: Introduction to Machine Learning. Morgan Kaufmann Publishers, Inc., London (1988)
Kubat, M.: An Introduction to Machine Learning. Springer, Heidelberg (2015). https://doi.org/10.1007/978-3-319-20010-1
Lichtenberg, G., Neidig, J.: Discrete event control of a pinball machine. In: Proceedings of 7th IFAC Workshop on Discrete Event Systems, Reims, France, pp 133–138 (2004)
Mnih, V., et al.: Human-level control through deep reinforcement learning. Nature 518, 529–533 (2015)
Sugiyama, M.: Statistical Reinforcement Learning - Modern Machine Learning Approaches. Taylor & Francis Group, LLC, Boca Raton (2015)
Sutton, R., Barto, A.: Reinforcement Learning: An Introduction, 2nd edn. MIT Press, Cambridge (2018)
Vicon Motion Systems: Vicon Tracker User Guide (2018). https://docs.vicon.com/display/Tracker37. Accessed January 2020
Acknowledgments
The authors thank the volunteer test persons who made themselves available for generating a database as a comparison for the results presented in this paper.
Further thanks are going to sponsors of IAI2020 Conference for their intellectual and financial support.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Alpen, M., Herzig, S., Horn, J. (2022). Towards Reinforcement Learning Control of an Electromechanical Pinball Machine. In: Karim, R., Ahmadi, A., Soleimanmeigouni, I., Kour, R., Rao, R. (eds) International Congress and Workshop on Industrial AI 2021. IAI 2021. Lecture Notes in Mechanical Engineering. Springer, Cham. https://doi.org/10.1007/978-3-030-93639-6_1
Download citation
DOI: https://doi.org/10.1007/978-3-030-93639-6_1
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-93638-9
Online ISBN: 978-3-030-93639-6
eBook Packages: EngineeringEngineering (R0)