Context identification for efficient multiple-model state estimation

Abstract

This paper presents an approach to accurate and scalable multiple-model state estimation for hybrid systems with intermittent, multi-modal dynamics. The approach consists of using discrete-state estimation to identify a system’s behavioral context and determine which motion models appropriately represent current dynamics, and which multiple-model filters are appropriate for state estimation. This improves the accuracy and scalability of conventional multiple-model state estimation. This approach is validated experimentally on a mobile robot that exhibits multi-modal dynamics.