Hey!

Welcome to Blip, home of the best original web series! When you’re done watching this episode, checkout some of our top shows or learn more about us!

×

Nick

Alergic - Non-Localizable Robustness - 3rd Nov 2010

Other Sharing Options

×
Embed
The embed code has been copied to your clipboard
Share
About this episode
Lessons from minimal bio-inspired systemsLONG-ABSTRACT: Theoretical discussions and computational models of bio-inspired embodied and situated agent...
Lessons from minimal bio-inspired systemsLONG-ABSTRACT: Theoretical discussions and computational models of bio-inspired embodied and situated agents are presented in this seminar capturing in simplified form the dynamical essence of robust and adaptive behaviour. The general problem of how dynamical coupling between internal control (brain), body, and environment are exploited in the generation of behaviour is particularly analyzed.This seminar proposes that the growing consensus about the importance of brain-body-environment couplings is still a minority view in several disciplines investigating robustness. These disciplines include cognitive and behavioural neuroscience, a good part of AI and robotics, and indeed several areas of biology. It is to be expected that even for those who agree on the view promoted in this seminar (i.e. the distribution of mechanisms promoting behavioural robustness), the full implications of it have not fully been drawn, and this talk attempts to address the proposed discussion from a bio-inspired systems context.Four experiments are introduced to support discussions using evolutionary algorithms based on Evolutionary Robotics (ER) methodology to generate the appropriate neural control. The first model evolves dynamically robust engagements for goal seeking in the presence of neural noise perturbations. The second model develops cognitive-behavioural dependencies for minimal-cognitive behaviour in dynamically limited agents. The third one evolves experience-dependent robust behaviour in one-legged agent walking.Finally, the last model shows functional dependencies in a mobile-object tracking task. These experiments include a series of perturbations (e.g. structural, sensorimotor, or mutational), or in the absence of them.Experimental results indicate that neural controls are not sufficient to generate robust behaviour in each case, suggesting the absence of internal control ‘ensuring’ robustness (a common believe in systems biology). The general observation is that coupling dynamics ‘forces’ evolution to behavioural robustness in whatever dynamical form evolution cares to come up with, but relying on behavioural mechanisms with certain dependence to brain, body, and environment dynamics. Experimental observations provide testable hypothesis that are likely to address in simple organisms in the biological realm, which has some implications for theoretical biology and artificial systems design. Less
01:04:00
Discover the best in original web series.© 2012 Blip Networks, Inc. All Rights Reserved.