Using Backprojections for Fine Motion Planning with Uncertainty - Robotics Institute Carnegie Mellon University

Using Backprojections for Fine Motion Planning with Uncertainty

Journal Article, International Journal of Robotics Research, Vol. 5, No. 1, pp. 19 - 45, March, 1986

Abstract

This paper outlines a method for planning motions in the presence of uncertainty. Tasks are modeled as geometrical goals in configuration space. The planning process consists of determining regions from which particular motions are guaranteed to reach a desired goal successfully. An algorithm is presented for backprojecting from desired goal states. The backprojection regions are computed by erecting constraints that geometrically capture the uncertainty in motion. The relationship of backprojections to goal recognizability is discussed within the formal framework of preimages. This relationship suggests a partitioning of desired goal states into recognizable goal states. Backprojections are actually performed from this partitioning.

BibTeX

@article{Erdmann-1986-15657,
author = {Michael Erdmann},
title = {Using Backprojections for Fine Motion Planning with Uncertainty},
journal = {International Journal of Robotics Research},
year = {1986},
month = {March},
volume = {5},
number = {1},
pages = {19 - 45},
}