Publication: Simple Agents Benefit Only From Simple Brains
Full text at PDC
Advisors (or tutors)
World Academy of Science, Engineering & Technology
In order to answer the general question: "What does a simple agent with a limited life-time require for constructing a useful representation of the environment?" we propose a robot platform including the simplest probabilistic sensory and motor layers. Then we use the platform as a test-bed for evaluation of the navigational capabilities of the robot with different "brains". We claim that a protocognitive behavior is not a consequence of highly sophisticated sensory-motor organs but instead emerges through an increment of the internal complexity and reutilization of the minimal sensory information. We show that the most fundamental robot element, the short-time memory, is essential in obstacle avoidance. However, in the simplest conditions of no obstacles the straightforward memory-less robot is usually superior. We also demonstrate how a low level action planning. involving essentially nonlinear dynamics, provides a considerable gain to the robot performance dynamically changing the robot strategy. Still, however, for very short life time the brainless robot is superior. Accordingly we suggest that small organisms (or agents) with short life-time does not require complex brains and even can benefit from simple brain-like (reflex) structures. To some extend this may mean that controlling blocks of modern robots are too complicated comparative to their life-time and mechanical abilities.
Conference of the World-Academy-of-Science-Engineering-and-Technology. Barcelona, SPAIN. OCT 22-24, 2006.
R.D. Beer, “Toward the evolution of dynamical neural networks for minimally cognitive behavior”, From animals to animats 4. In Maas P., Mataric M., Meyer J., Pollack J., and Wilson S. (Eds.). Proceedings of the Fourth International Conference on Simulation of Adaptive Behavior MIT Press, 1996, pp. 421–429. R.D. Beer, “The dynamics of active categorical perception in an evolved model agent”. Adaptive Behavior, vol. 11, no.4, pp. 209–243, 2003. D. Kortenkamp and T. Weymouth, “Topological mapping for mobile robots using a combination of sonar and vision sensing”, Proc of the AI, pp. 979–984, 1994. U. Ulrich and J. Borenstein, “Reliable obstacle avoidance for fast mobile robots”, IEEE Int. Conf. on Robotics and Automation, pp. 1572–1577, 1998. K. Arras, T. Tomaris, B. Jensen, and R. Siegwart, “Multisensor on the fly localization: Precision and reliability for applications”, Robotics and Autonomous Systems, vol. 34, pp. 131–143, 2001. S. Thrun, “Probabilistic algorithms in robotics”, AI Magazine vol. 21, no. 4, pp. 93–109, 2000. A. Atrash and S. Koening, “Probabilistic Planning for Behavior-Based Robot”. Proc Flairs Conference, pp. 531–535, 2001. S. Engelson and D. McDermott, “Error correction in mobile robot map learning”, Proc of the 1992 IEEE Int. Conf. on Robotics and Automation, pp. 2555–2560, 1992. R. Jaulmes, J. Pineau, and D. Precup, “Probabilistic robot planning under model uncertainty: an active learning approach”. NIPS Workshop on Machine Learning Based Robotics in Unstructured Environments, 2005. F. Atteneave, “Some informational aspect of visual perception”, Psychol Rev, vol. 61, pp. 183–193, 1954. H. Barlow, Sensory communication. Cambridge, Massachusetts: MIT Press, 1961. J. Atick and N. Redlich, “Towards a theory of early visual processing”. Neural Comput, vol. 2, pp. 308–320, 1990. J. Atick, “Could information theory provide an ecological theory of sensory processing?”. Network, vol. 3, pp. 213-251, 1992. J. Atick and W. Bialek, Princeton Lectures on Biophysics, W. Bialek. World Scientific, Singapore, 1992. J. Baddeley and N. Weinberg, “Induction of a physiological memory in the cerebral cortex by stimulation of the nucleus basalis”, Proc. Natl. Acad. Sci. USA, vol. 93, pp. 11219–11224, 1996. R. Brooks, “A robust layered control system for a mobile robot”, IEEE J. Rob. Autom, vol. 2,pp. 14–23, 1986. A. Brooks, “Hardware retargetable distributed layered architecture for mobile robot control”. Proc IEEE Robotics and Automation, Raleigh, NC, pp. 106–110, 1987.