Localization for multi-axle train configured CFMMRs

Abstract

This paper presents a data fusion structure based on comparing geometric configurations of serial connected multi-axle compliant framed robots. Data sources include global odometry derived sources and a novel strain-measurement based relative posture sensor (RPS). Geometric methods are used because stochastic data fusion, developed from prior research, was erroneous when applied to more generalized multi-axle configurations. Our results show an excellent response predicting expected configurations and a reasonable response with un-expected configurations.