Sciweavers

797 search results - page 4 / 160
» Timed Control with Partial Observability
Sort
View
DSN
2006
IEEE
14 years 9 days ago
Automatic Recovery Using Bounded Partially Observable Markov Decision Processes
This paper provides a technique, based on partially observable Markov decision processes (POMDPs), for building automatic recovery controllers to guide distributed system recovery...
Kaustubh R. Joshi, William H. Sanders, Matti A. Hi...
ICRA
2007
IEEE
126views Robotics» more  ICRA 2007»
14 years 17 days ago
A formal framework for robot learning and control under model uncertainty
— While the Partially Observable Markov Decision Process (POMDP) provides a formal framework for the problem of robot control under uncertainty, it typically assumes a known and ...
Robin Jaulmes, Joelle Pineau, Doina Precup
MCSS
2002
Springer
13 years 6 months ago
Observability and Forward-Backward Observability of Discrete-Time Nonlinear Systems
In this paper, we study the observability properties of nonlinear discrete time systems. Two types of contributions are given. First, we present observability criteria in terms of...
Francesca Albertini, Domenico D'Alessandro
AAAI
2007
13 years 8 months ago
On the Partial Observability of Temporal Uncertainty
We explore a means to both model and reason about partial observability within the scope of constraintbased temporal reasoning. Prior studies of uncertainty in Temporal CSPs have ...
Michael D. Moffitt
AAAI
1997
13 years 7 months ago
Incremental Methods for Computing Bounds in Partially Observable Markov Decision Processes
Partially observable Markov decision processes (POMDPs) allow one to model complex dynamic decision or control problems that include both action outcome uncertainty and imperfect ...
Milos Hauskrecht