This dissertation explores and analyzes the performance of several Bayesian anytime inference algorithms for dynamic influence diagrams. These algorithms are compared on the On-Line Maintenance Agent testbed, a softwar ...