Thesis

  • Hyperfiltering for Stochastic systems
    James C. Davidson
    Master's Thesis
    University of Illinois, Urbana, Illinois, 2007
    Abstract :: BibTex :: Download PDF

    Abstract: Information-feedback control schemes (more specifically, sensor-based control schemes) select an action at each stage based on the sensory data provided at that stage. Since it is impossible to know future sensor readings in advance, predicting the future behavior of a system becomes necessary. Hyperfiltering is a sequential method that enables probabilistic evaluation of future system performance in the face of this uncertainty. Rather than evaluating individual sample paths or relying on point estimates of state, hyperfiltering maintains at each stage an approximation of the full probability density function over the belief space (i.e., the space of possible posterior densities for the state estimate).

    @mastersthesis{Dav07_msthesis,
    	title = {Hyperfiltering for stochastic systems},
    	author = {James C. Davidson},
    	school = {University of Illinois at Urbana-Champaign},
    	year = {2007}
    }
    				

Conference Proceedings

  • A Sampling Hyperbelief Optimization Techinque for Stochastic Systems
    James C. Davidson and Seth A. Hutchinson
    in the International Workshop on the Algorithmic Foundations of Robotic (WAFR) 2008
    Abstract :: BibTex :: Download PDF

    In this paper we propose an anytime algorithm for determining nearly optimal policies for total cost and finite time horizon partially observed Markov decision processes (POMDPs) using a sampling-based approach. The proposed technique, sampling hyperbelief optimization technique (SHOT), attempts to exploit the notion that small changes in a policy have little impact on the quality of the solution except at a small set of critical points. The result is a technique to represent POMDPs independent of the initial conditions and the particular cost function, so that the initial conditions and the cost function may vary without having to reperform the majority of the computational analysis.

    @inproceedings{DavHut08_shot,
    	title = {A Sampling Hyperbelief Optimization Techinque for Stochastic Systems},
    	author = {James C. Davidson and Seth A. Hutchinson},
    	booktitle = {Proceedings of the International Workshop on the Algorithmic Foundations of Robotics},
    	year = {2008}
    }
    				
  • Hyper-particle Filtering for stochastic systems
    James C. Davidson and Seth A. Hutchinson
    in the IEEE Interational Conference on Robotics, 2008
    Abstract :: BibTex :: Download PDF :: Web Link

    Abstract: Information-feedback control schemes (more specifically, sensor-based control schemes) select an action at each stage based on the sensory data provided at that stage. Since it is impossible to know future sensor readings in advance, predicting the future behavior of a system becomes difficult. Hyper-particle filtering is a sequential computational scheme that enables probabilistic evaluation of future system performance in the face of this uncertainty. Rather than evaluating individual sample paths or relying on point estimates of state, hyper-particle filtering maintains at each stage an approximation of the full probability density function over the belief space (i.e., the space of possible posterior densities for the state estimate). By applying hyper-particle filtering, control policies can be more more accurately assessed and can be evaluated from one stage to the next. These aspects of hyper-particle filtering may prove to be useful when determining policies, not just when evaluating them.

    @inproceedings{DavHut08_part,
    	title     = {Hyper-particle filtering for stochastic systems},
    	author    = {James C. Davidson AND Seth A. Hutchinson},
    	booktitle = {Proceedings of the IEEE International Conference on Robotics},
    	pages     = {2770-2777},
    	year      = {2008},
    	month     = {May}
    }
    				
  • Recognition of traversable areas for mobile robotic navigation in outdoor environments
    James C. Davidson and Seth A. Hutchinson
    in the IEEE/RSJ International Conference on Intelligent Robotics and Systems, 2003
    Abstract :: BibTex :: Download PDF :: Web Link

    Abstract: In this paper we consider the problem of automatically determining whether regions in an outdoor environment can be traversed by a mobile robot. We propose a two-level classifier that uses data from a single color image to make this determination. At the low level, we have implemented three classifiers based on color histograms, directional filters and local binary patterns. The outputs of these low level classifiers are combined using a voting scheme that weights the results of each classifier using an estimate of its error probability. We present results from a large number of trials using a database of representative images acquired in real outdoor environments.

    @INPROCEEDINGS{DavHut03,
    	AUTHOR    = {James C. Davidson and Seth A. Hutchinson},
    	TITLE     = {Recognition of traversable areas for mobile robotic navigation in outdoor environments},
    	BOOKTITLE = {IEEE/RSJ International Conference on Intelligent Robotics
    	and Systems},
    	PAGES     = {297-304},
    	YEAR      = {2003},
    	MONTH     = {October}
    }