Learning environmental features for pose estimation

We present a method for learning a set of environmental features which are useful for pose estimation. The landmark learning mechanism is designed to be applicable to a wide range of environments, and generalized for differen t sensing modalities. In the context of computer vision, each landmark is d etected as a local extremum of a measure of distinctiveness and represented by an appearance-based encoding which is exploited for matching. The set o f obtained landmarks can be parameterized and then evaluated in terms of th eir utility for the task at hand. The method is used to motivate a general approach to task-oriented sensor fusion. We present experimental evidence t hat demonstrates the utility of the method. (C) 2001 Elsevier Science B.V. All rights reserved.