ICML-98 Submission #194

	       Learning To Locate An Object in 3D Space
		   from A Sequence Of Camera Images


			 Dimitris Margaritis
			  Sebastian Thrun

		      Dept. of Computer Science
		      Carnegie Mellon University
			   5000 Forbes Ave.
			 Pittsburgh, PA 15213


			       Abstract

  This paper addresses the problem of determining an object's 3D
  location from a stream of camera images recorded by mobile robot.
  The approach presented here allows people to ``train'' robots to
  recognize specific objects, by presenting it examples of the object
  to be recognized. A decision tree method is used to learn
  significant features of the target object from individual camera
  snapshots.  Individual estimates are integrated over time using
  Bayes rule, into a probabilistic 3D model of the robot's
  environment.  Experimental results illustrate that the method
  enables a mobile robot to robustly estimate the 3D location of
  objects from multiple camera images.

Keywords: Mobile Robotics, Decision Trees, Probabilistic Reasoning

Contact author:

Dimitris Margaritis (dmarg+@cs.cmu.edu)
office: (412) 268-6767
lab:    (412) 268-3837