Controlling Contractors with Monads for Hybrid Dynamical Systems
Gilles Chabert, Rémi Douence · 2010
Physical systems with intelligent behaviors result from inter-actions of different fields: sensor networks, robotics, optimization, reasoning, etc. Rooted in this philosophy of interdisciplinary, this paper makes a connexion between hybrid dynamical systems, interval-based constraint propagation and functional programming. It shows how to build a monadic program in Haskell to control contractors (constraint propagators) for the state estimation of multi-model (hy- brid) dynamical systems, subject to partial and uncertain measurements. The example of system taken here is an elevator that can either be moving upward, downward or stopped. The altitude is measured directly and the estimation problem is simply to track its motion. The purpose of the Haskell library is to offer both a high-level and flexible framework for building propagation strategies based on user knowledge or user requirements.