BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Incremental Neuroevolution of Reactive and Deliberative 3D Agents 
 - Adam Stanton\, Keele University
DTSTART:20150930T120000Z
DTEND:20150930T130000Z
UID:TALK61170@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Following earlier work on the neuroevolution of deliberative b
 ehaviour to solve increasingly challenging tasks in a two- dimensional dyn
 amic world\, in this talk I will present the implementation and results of
  extending the original system to a three-dimensional rigid body simulatio
 n.\n\nThe 3D physically based setting requires that a successful agent con
 tinually and deliberately adjust its gait\, turning and other motor contro
 l over the many stages and sub-stages of these tasks\, within its individu
 al evaluation. Achieving such complex interplay between motor control and 
 deliberative control\, within a neuroevolutionary framework\, is the focus
  of this work. To this end\, a novel neural architecture is presented and 
 an incremental evolutionary approach used to bootstrap the locomotive beha
 viour of the agents. Agent morphology is fixed as a quadruped with three d
 egrees of freedom per limb. Agent populations have no initial knowledge of
  the problem domain\, and evolve to move around and then solve progressive
 ly more difficult challenges in the environment using a tournament-based c
 o-evolutionary algorithm. The results demonstrate not only success at the 
 tasks but also a variety of intricate lifelike behaviours being used\, sep
 arately and in combination\, to achieve this success. Given the problem-ag
 nostic controller architecture\, these results indicate a potential for di
 scovering yet more advanced behaviours in yet more complex environments.\n
 Abstract not available
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR