An Approach to Reasoning About Continuous Change for Applications in Planning

Thomas Dean, Greg Siegle

There are many planning applications that require an agent to coordinate its activities with processes that change continuously over time. Several proposals have been made for combining a temporal logic of time with the differential and integral calculus to provide a hybrid calculus suitable for planning applications. We take one proposal and explore some of the issues involved in implementing a practical system that derives conclusions consistent with such a hybrid calculus. Models for real-valued parameters are specified as systems of ordinary differential equations, and constructs are provided for reasoning about how these models change over time. For planning problems that require projecting the consequences of a set of events from a set of initial conditions and causal rules, a combination of numerical approximation and symbolic math routines and a simple default reasoning strategy provide for an efficient inference engine.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.