Abstract
Collective autonomic systems (CAS) are distributed collections of agents that collaborate to achieve the system’s goals but autonomously adapt their behavior. We present the teacher/student architecture for locally coordinated distributed learning and show that in certain scenarios the performance of a swarm using teacher/student learning can be significantly better than that of agents learning individually. Teacher/student learning serves as foundation for the continuous collaboration (CC) development approach. We introduce CC, relate it to the EDLC, a life cycle model for CAS, and show that CC embodies many of the principles proposed for developing CAS.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
Taking into account the different times at which students exchange information with teachers, the knowledge and strategies the students share are typically similar, not identical. This does not change the gist of the following discussion.
- 2.
The careful reader may observe that the single robot takes only approximately 6 times as long as the swarm to reach its maximal performance, not more than 10 times as might be expected. This is an artifact of our learning schedule which learns only at the end of each episode, so that the single agent performs many more iterations of the DP algorithm before it reaches its maximum performance than the DP-learner and thus better exploits the data it has available. This means that the single agent can focus a larger percentage of its exploration on promising parts of the graph, thereby negating the advantages that the swarm has over a single learner. However, a swarm of 10 single learners would use 10 times the computational resources of a swarm with a DP-learner, which would justify running the DP-learner 10 times as frequently with corresponding improvements to the swarm’s performance.
- 3.
Between episodes 30 and 50 the random modifications result in a graph in which some of the routes computed by the non-learning teachers are viable, therefore the performance is slightly better than in the other episodes in which the graph is damaged.
References
Andre, D.: Programmable reinforcement learning agents. Ph.D. thesis, University of California at Berkeley (2003)
Ay, N., Der, R., Prokopenko, M.: Guided self-organization: perception-action loops of embodied systems. Theory Biosci. 131(3), 125–127 (2012)
Belzner, L., Hölzl, M., Koch, N., Wirsing, M.: Collective autonomic systems: towards engineering principles and their foundations, July 2016
Cheng, B., et al.: Software engineering for self-adaptive systems: a research roadmap. In: Cheng, B., de Lemos, R., Giese, H., Inverardi, P., Magee, J. (eds.) Software Engineering for Self-Adaptive Systems. LNCS, vol. 5525, pp. 1–26. Springer, Heidelberg (2009). doi:10.1007/978-3-642-02161-9_1
Claus, C., Boutilier, C.: The dynamics of reinforcement learning in cooperative multiagent systems. In: Proceedings of the 15th National/Tenth Conference on AI/Innovative Applications of AI, AAAI 1998/IAAI 1998, pp. 746–752. AAAI (1998)
Colombo, A., Fontanelli, D., Legay, A., Palopoli, L., Sedwards, S.: Efficient customisable dynamic motion planning for assistive robots in complex human environments. J. Ambient Intell. Smart Environ. 7(5), 617–634 (2015)
Fagin, R., Moses, Y., Vardi, M., Halpern, J.: Reasoning About Knowledge. MIT Press, Cambridge (2003)
Ghallab, M., Nau, D.S., Traverso, P.: Automated Planning - Theory and Practice. Elsevier, Amsterdam (2004)
Hölzl, M., Gabor, T.: Continuous collaboration: a case study on the development of an adaptive cyber-physical system. In: Proceedings of the 1st International Workshop on Software Engineering for Smart Cyber-Physical Systems (SEsCPS). IEEE (2015)
Hölzl, M., Gabor, T.: Reasoning and learning for awareness and adaptation. In: Wirsing et al. [29]
Hölzl, M., Koch, N., Puviani, M., Wirsing, M., Zambonelli, F.: The ensemble development life cycle and best practices for collective autonomic systems. In: Wirsing et al. [29]
Karafotias, G., Haasdijk, E., Eiben, A.E.: An algorithm for distributed on-line, on-board evolutionary robotics. In: Proceedings of the 13th Annual Conference on Genetic and Evolutionary Computation, GECCO 2011, pp. 171–178. ACM, New York (2011)
Kephart, J.O., Chess, D.M.: The vision of autonomic computing. Computer 36(1), 41–50 (2003)
Marzinotto, A., Colledanchise, M., Smith, C., Ögren, P.: Towards a unified behavior trees framework for robot control. In: 2014 IEEE International Conference on Robotics and Automation, ICRA 2014, Hong Kong, pp. 5420–5427. IEEE (2014)
Millington, I., Funge, J.: Artificial Intelligence for Games, 2nd edn. Morgan Kaufmann, San Francisco (2009)
Object Management Group: UML Specifications. http://www.omg.org/spec/. Accessed 26 Feb 2015
Ogren, P.: Increasing modularity of UAV control systems using computer game behavior trees. In: AIAA Guidance, Navigation and Control Conference, Minneapolis, Minnesota, pp. 13–16 (2012)
Shalev-Shwartz, S., Ben-David, S.: Understanding Machine Learning: From Theory to Algorithms. Cambridge University Press, New York (2014)
Shoham, Y., Leyton-Brown, K.: Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations. Cambridge University Press, New York (2008)
Sigmund, K.: A survey of replicator equations. In: Casti, J.L., Karlqvist, A. (eds.) Complexity, Language, and Life: Mathematical Approaches. Biomathematics, vol. 16, pp. 88–104. Springer, Heidelberg (1986)
Sutton, R.S., Barto, A.G.: Reinforcement Learning. MIT Press, Cambridge (1998)
Thrun, S., Burgard, W., Fox, D.: Probabilistic Robotics. MIT Press, Cambridge (2005)
Vapnik, V.: The Nature of Statistical Learning Theory. Information Science and Statistics. Springer, New York (2013)
Vapnik, V.N.: Statistical Learning Theory. Wiley-Interscience, New York (1998)
Watson, R.A., Ficici, S.G., Pollack, J.B.: Embodied evolution: distributing an evolutionary algorithm in a population of robots. Robot. Auton. Syst. 39(1), 1–18 (2002)
Weiss, G. (ed.): Multiagent Systems, 2nd edn. MIT Press, Cambridge (2013)
Wiegand, R.P.: An analysis of cooperative coevolutionary algorithms. Ph.D. thesis, George Mason University (2003)
Wiering, M., van Otterlo, M.: Reinforcement Learning: State-of-the-Art. Adaptation, Learning, and Optimization, vol. 12. Springer, Heidelberg (2012)
Wirsing, M., Hölzl, M., Koch, N., Mayer, P. (eds.): Software Engineering for Collective Autonomic Systems: Results of the ASCENS Project. LNCS, vol. 8998. Springer, Heidelberg (2015)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing AG
About this chapter
Cite this chapter
Hölzl, M., Gabor, T. (2016). Continuous Collaboration for Changing Environments. In: Steffen, B. (eds) Transactions on Foundations for Mastering Change I. Lecture Notes in Computer Science(), vol 9960. Springer, Cham. https://doi.org/10.1007/978-3-319-46508-1_11
Download citation
DOI: https://doi.org/10.1007/978-3-319-46508-1_11
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-46507-4
Online ISBN: 978-3-319-46508-1
eBook Packages: Computer ScienceComputer Science (R0)