Abstract
This paper describes the development of an Artificial Neural Network (ANN) architecture that is capable to implement reactive behavior in Autonomous Agents (AAs). We make considerations about biological paradigms, as evolutionary mechanisms and animals' behaviors, trying to find solutions that, once applied to the development of artificial devices, provide more robust and useful AAs to operate in the real world. To have higher survival chances, life beings must have developed more complex behaviors, as reactive and internally motivated, in which the agent's action must depend on the past history of sensory values in order to be effective. This work shows that reactive behaviors can be described through Finite State Machines (FSMs) that requires a recurrent neural network architecture to be implemented, in order to insert dynamics and memory in the system. After the description of such network architecture and a biologically inspired learning algorithm, we make some experiments. Finally, conclusions for future work are given.
This is a preview of subscription content, log in via an institution.
Preview
Unable to display preview. Download preview PDF.
References
Anderson, T. L. and Donath, M. Animal Behavior as a Paradigm for Developing Robot Autonomy. In. Maes, P. (ed.) Designing Autonomous Agents. Theory and Practice from Biology to Engineering and Back. MIT Press. 1991. p. 145–168.
Angeline, P. J. Evolutionary Algorithms and Emergent Intelligence, Ph.D. Thesis — Ohio State University. USA. 1993. pp. 172.
Barreto, J. M. Connectionism in Problem Solving, Titular Professor contest Monograph, UFSC, Florianópolis, March 1996 (in Portuguese).
Beer, R. D. et al. A Biological Perspective on Autonomous Agent Design. In. Maes, P. (ed.) Designing Autonomous Agents. Theory and Practice from Biology to Engineering and Back. MIT Press. 1991. p. 169–186.
Brooks, R. A. Elephants Don't Play Chess. In. Maes, P. (ed.) Designing Autonomous Agents. Theory and Practice from Biology to Engineering and Back. MIT Press. 1991. p. 3–15.
Freeman, J. A. and Skapura, D. M. Neural Networks, Algorithms, Applications and Programming Techniques. Addison-Wesley. 1991. pp 401.
Kaelbling, L. P. Learning in Embedded Systems. MIT Press. 1993. pp 176.
Kuperstein, M. Neural Networks that Create their Own Goals with Growth Cycles. In Proc. of the World Congress on Neural Networks. Portland, Oregon. 1993. p.111–122. Vol. 3.
Lin, L. Programming Robots Using Reinforcement Learning and Teaching. In. Proc. of the Ninth Nat. Conf. on Artif. Intel. USA. 1991. p.781–786.
Maes, P. Guest Editorial. In Maes, P. (ed.) Designing Autonomous Agents. Theory and Practice from Biology to Engineering and Back. MIT Press. 1991. p.1–2.
Mahadevan, S. and Connell, J. Automatic Programming of Behavior-based Robots using Reinforcement Learning. In. Proc. of the Ninth Nat. Conf. on Artif. Intel. USA. 1991. p. 768–773.
Manning, A. An introduction to Animal Behavior. 3rd ed. Addison-Wesley, Massachusetts, 1979.
Piggott, P. and Sattar, S. Reinforcement Learning of Iterative Behaviour with Multiple Sensors. Journal of Applied Intel., 4, 1994. p.351–365.
Vaario, J. An Emergent Modeling Method for Artificial Neural Networks. Doctor Dissertation of Engineering-The Univ. of Tokyo. Japan. 1993. pp. 216.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1996 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Roisenberg, M., Barreto, J.M., Azevedo, F.M. (1996). Biological inspirations in neural network implementations of Autonomous Agents. In: Borges, D.L., Kaestner, C.A.A. (eds) Advances in Artificial Intelligence. SBIA 1996. Lecture Notes in Computer Science, vol 1159. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-61859-7_22
Download citation
DOI: https://doi.org/10.1007/3-540-61859-7_22
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-61859-1
Online ISBN: 978-3-540-70742-4
eBook Packages: Springer Book Archive