<

Statisztikus Fizikai Szemináriumok
Seminars in Statistical Physics

A szemináriumokat az Elméleti Fizikai Tanszék (1117 Budapest, Pázmány Péter sétány 1/A, azaz Északi tömb) keretében szervezzük. Az előadások ebben a félévben az É 2.54-es teremben (Novobátzky Károly terem) kerülnek megrendezésre. Szokásosan szerdán de. 11-kor kezdünk. Ha ettől eltérő időben vagy más teremben tartjuk az előadást, azt külön jelezzük. Az előadás címét általában annak nyelvén hirdetjük meg.


The seminars are held at the Institute for Theoretical Physics, Eötvös University (1117 Budapest, Pázmány Péter sétány 1/A, i.e. Northern Block near Petőfi bridge, western bridgehead). This semester the seminars are held in room É 2.54 (Novobátzky Károly Hall). Usually the seminar is on Wednesday 11 am. If we deviate from the standard times or location, there will be an extra warning. We announce generally the title of the lecture in the language the speaker will use.

12th April 2017

Sándor Bulcsú

Institute for Theoretical Physics, Goethe University

A dynamical systems approach to robotics: the role of attractors in locomotion

The ultimate goal in robotic locomotion is designing and constructing robots which are able to navigate and transport themselves autonomously from place to place. Robots, situated in an environment, need hence to interact with the external world via their own body and incorporate the feedback in a closed-loop control scheme. Following the recently emerged "robophysics" approach to locomotion, we investigate simple rolling robots within the LPZRobots simulation environment. The "brain" of the considered robots consists of a single, respectively three propriceptual neurons, receiving sensory information only about the position of the actuators. The proposed minimal control mechanism allows for a self-organized generation of motion patterns as stable limit cycles or chaotic attractors in the overarching phase space of the brain, body, and environment. The interactions with obstacles or other agents can hence be interpreted, in terms of dynamical systems theory, as switching between coexisting attractors or attractor ruins. We argue that this dynamical systems approach may contribute both to the general understanding of locomotion and to the design of learnable autonomous agents.

Download announcement in PDF

Korábbi előadások: 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
2010 2011 2012 2013 2014 2015 2016 2017

Keresés az előadások között:
(powered by Google)

Page maintained by Norbert Barankai --- Barankai Norbert tartja karban ezt az oldalt.