ITEE seminar: Mr. David Ball, 11.00AM, Fri 09 May 2003

Predicting an Opponent's Future in a Highly Dynamic, Multi-Agent and Inaccessible Environment.

Speaker: Mr. David Ball, ITEE

When: 11.00AM, Friday 09 May 2003

Venue: 78-420

Host: Dr. Gordon Wyeth

Abstract:

  A proposal is presented for predicting the future behaviour of
  opponent agents in a competitive, highly dynamic, multi-agent and
  inaccessible environment, namely RoboCup. An existing planning
  system will be adapted to use these predictions to better plan for
  the future. This research is broken into three tasks; classification
  of behaviours, modelling and prediction of behaviours and
  integration of the predictions into the existing planning
  system. The focus is on learning and predicting the behaviour of
  individual agents, not an overall team strategy. A probabilistic
  approach is taken to dealing with the uncertainty in the
  observations and with representing the uncertainty in the prediction
  of the behaviours. Learning of the opponent's behaviours is
  iterative and gradual.  The environment is inaccessible in the sense
  that it is not possible to directly observe the opponent's
  behaviours. Instead they must be reasoned about. A classification
  system using a Naïve Bayesian Network will determine the opponent's
  current behaviour. The modelling system will use 2D discrete and
  continuous probability distributions to model the manner with which
  the opponents perform their behaviours. The prediction system will
  use the results from the classification system and the modelling
  system to generate a 2D probability distribution of the opponent's
  predicted future. Finally these predictions and an estimate of their
  accuracy will be integrated into the existing planning system.  A
  research platform consisting of five real robots (the RoboRoos) has
  been built and tested. Their intelligence systems including a
  simulator and a playback/logging system has been written to be
  robust, expandable and easy to use. The research platform has been
  tested with promising initial results.

Biography:

  David completed a bachelors degree in Computer Systems Engineering
  (Mechatronics Minor) at the University of Queensland in 2001. He has
  represented the University of Queensland internationally twice at
  RoboCup competitions (Seattle, USA and Fukuoka, Japan). He is
  currently researching opponent behaviour prediction in a RoboCup
  environment (in a PhD program)

Type:

Ph.D Confirmation

Contact:

Dr. Gordon Wyeth, seminar host (wyeth@itee.uq.edu.au)
or Guido Governatori (ITEE seminar co-ordinator) (guido@itee.uq.edu.au)

ITEE seminar web page: http://www.itee.uq.edu.au/~seminar

[All seminars]